From 7b8472c6e9f7d66ee604cf7ee3c6a1ef6ab5dc59 Mon Sep 17 00:00:00 2001 From: Nicholas Clark Date: Tue, 28 Mar 2023 19:45:22 +1000 Subject: [PATCH] harmonise parameter names; begin adding support for other exponential families --- R/add_base_dgam_lines.R | 40 +- R/add_poisson_lines.R | 2 +- R/add_stan_data.R | 16 +- R/add_trend_lines.R | 120 ++-- R/add_tweedie_lines.R | 8 +- R/eval_mvgam.R | 172 +++--- R/families.R | 123 ++++ R/forecast.mvgam.R | 58 +- R/get_monitor_pars.R | 17 +- R/get_mvgam_priors.R | 20 +- R/get_mvgam_resids.R | 8 +- R/mvgam.R | 132 ++-- R/mvgam_predict.R | 48 +- R/mvgam_setup.R | 11 +- R/pfilter_mvgam_fc.R | 26 +- R/pfilter_mvgam_init.R | 144 +++-- R/pfilter_mvgam_smooth.R | 103 ++- R/predict.mvgam.R | 37 +- R/print.mvgam.R | 12 +- R/summary.mvgam.R | 40 +- R/vectorise_stan_lik.R | 14 +- docs/README-unnamed-chunk-11-1.png | Bin 19291 -> 17071 bytes docs/README-unnamed-chunk-15-1.png | Bin 8047 -> 7953 bytes docs/README-unnamed-chunk-16-1.png | Bin 10911 -> 10868 bytes docs/README-unnamed-chunk-17-1.png | Bin 9477 -> 9453 bytes docs/README-unnamed-chunk-18-1.png | Bin 7069 -> 7292 bytes docs/README-unnamed-chunk-19-1.png | Bin 9423 -> 9078 bytes docs/README-unnamed-chunk-21-1.png | Bin 16090 -> 15747 bytes docs/README-unnamed-chunk-22-1.png | Bin 23890 -> 23402 bytes docs/README-unnamed-chunk-23-1.png | Bin 9335 -> 9328 bytes docs/README-unnamed-chunk-24-1.png | Bin 19252 -> 17059 bytes docs/README-unnamed-chunk-24-2.png | Bin 21873 -> 22623 bytes docs/README-unnamed-chunk-25-1.png | Bin 16500 -> 16260 bytes docs/README-unnamed-chunk-26-1.png | Bin 13407 -> 13586 bytes docs/README-unnamed-chunk-27-1.png | Bin 22212 -> 22406 bytes docs/README-unnamed-chunk-28-1.png | Bin 8459 -> 6771 bytes docs/README-unnamed-chunk-29-1.png | Bin 8619 -> 8934 bytes docs/README-unnamed-chunk-30-1.png | Bin 9941 -> 10090 bytes docs/README-unnamed-chunk-31-1.png | Bin 8890 -> 9602 bytes docs/README-unnamed-chunk-32-1.png | Bin 14925 -> 14910 bytes docs/README-unnamed-chunk-33-1.png | Bin 19005 -> 19126 bytes docs/README-unnamed-chunk-8-1.png | Bin 21926 -> 17097 bytes docs/README.Rmd | 6 +- docs/README.md | 153 +++-- man/get_monitor_pars.Rd | 2 +- .../Case studies/mvgam_case_study1.Rmd | 13 +- .../Case studies/mvgam_case_study1.html | 375 ++++++----- .../Case studies/mvgam_case_study3.Rmd | 16 +- .../Case studies/mvgam_case_study3.html | 584 +++++++++--------- .../rpubs.com/rpubs/Publish Document.dcf | 4 +- .../rpubs.com/rpubs/Publish Document.dcf | 4 +- 51 files changed, 1204 insertions(+), 1104 deletions(-) create mode 100644 R/families.R diff --git a/R/add_base_dgam_lines.R b/R/add_base_dgam_lines.R index c569b8e3..3d4c2d85 100644 --- a/R/add_base_dgam_lines.R +++ b/R/add_base_dgam_lines.R @@ -223,23 +223,23 @@ add_base_dgam_lines = function(use_lv, stan = FALSE, offset = FALSE){ } for (j in 1:n_lv) { - LV_raw[2, j] ~ dnorm(phi[j] + ar1[j]*LV_raw[1, j], penalty[j]) + LV_raw[2, j] ~ dnorm(drift[j] + ar1[j]*LV_raw[1, j], penalty[j]) } for (j in 1:n_lv) { - LV_raw[3, j] ~ dnorm(phi[j] + ar1[j]*LV_raw[2, j] + ar2[j]*LV_raw[1, j], penalty[j]) + LV_raw[3, j] ~ dnorm(drift[j] + ar1[j]*LV_raw[2, j] + ar2[j]*LV_raw[1, j], penalty[j]) } for (i in 4:n) { for (j in 1:n_lv) { - LV_raw[i, j] ~ dnorm(phi[j] + ar1[j]*LV_raw[i - 1, j] + + LV_raw[i, j] ~ dnorm(drift[j] + ar1[j]*LV_raw[i - 1, j] + ar2[j]*LV_raw[i - 2, j] + ar3[j]*LV_raw[i - 3, j], penalty[j]) } } ## AR components for (s in 1:n_lv) { - phi[s] ~ dnorm(0, 10) + drift[s] ~ dnorm(0, 10) ar1[s] ~ dnorm(0, 10) ar2[s] ~ dnorm(0, 10) ar3[s] ~ dnorm(0, 10) @@ -318,9 +318,9 @@ add_base_dgam_lines = function(use_lv, stan = FALSE, offset = FALSE){ ## likelihood functions for (i in 1:n) { for (s in 1:n_series) { - y[i, s] ~ dnegbin(rate[i, s], r[s])T(, upper_bound[s]); - rate[i, s] <- ifelse((r[s] / (r[s] + mus[i, s])) < min_eps, min_eps, - (r[s] / (r[s] + mus[i, s]))) + y[i, s] ~ dnegbin(rate[i, s], phi[s])T(, upper_bound[s]); + rate[i, s] <- ifelse((phi[s] / (phi[s] + mus[i, s])) < min_eps, min_eps, + (phi[s] / (phi[s] + mus[i, s]))) } } @@ -328,14 +328,14 @@ add_base_dgam_lines = function(use_lv, stan = FALSE, offset = FALSE){ ## where the likelihood reduces to a 'base' model (Poisson) unless ## the data support overdispersion for (s in 1:n_series) { - r[s] <- 1 / r_inv[s] - r_inv[s] ~ dexp(5) + phi[s] <- 1 / phi_inv[s] + phi_inv[s] ~ dexp(5) } ## posterior predictions for (i in 1:n) { for (s in 1:n_series) { - ypred[i, s] ~ dnegbin(rate[i, s], r)T(, upper_bound[s]) + ypred[i, s] ~ dnegbin(rate[i, s], phi[s])T(, upper_bound[s]) } } @@ -361,22 +361,22 @@ add_base_dgam_lines = function(use_lv, stan = FALSE, offset = FALSE){ } for (s in 1:n_series) { - trend[2, s] ~ dnorm(phi[s] + ar1[s]*trend[1, s], tau[s]) + trend[2, s] ~ dnorm(drift[s] + ar1[s]*trend[1, s], tau[s]) } for (s in 1:n_series) { - trend[3, s] ~ dnorm(phi[s] + ar1[s]*trend[2, s] + ar2[s]*trend[1, s], tau[s]) + trend[3, s] ~ dnorm(drift[s] + ar1[s]*trend[2, s] + ar2[s]*trend[1, s], tau[s]) } for (i in 4:n) { for (s in 1:n_series){ - trend[i, s] ~ dnorm(phi[s] + ar1[s]*trend[i - 1, s] + ar2[s]*trend[i - 2, s] + ar3[s]*trend[i - 3, s], tau[s]) + trend[i, s] ~ dnorm(drift[s] + ar1[s]*trend[i - 1, s] + ar2[s]*trend[i - 2, s] + ar3[s]*trend[i - 3, s], tau[s]) } } ## AR components for (s in 1:n_series){ - phi[s] ~ dnorm(0, 10) + drift[s] ~ dnorm(0, 10) ar1[s] ~ dnorm(0, 10) ar2[s] ~ dnorm(0, 10) ar3[s] ~ dnorm(0, 10) @@ -387,9 +387,9 @@ add_base_dgam_lines = function(use_lv, stan = FALSE, offset = FALSE){ ## likelihood functions for (i in 1:n) { for (s in 1:n_series) { - y[i, s] ~ dnegbin(rate[i, s], r[s])T(, upper_bound[s]); - rate[i, s] <- ifelse((r[s] / (r[s] + mus[i, s])) < min_eps, min_eps, - (r[s] / (r[s] + mus[i, s]))) + y[i, s] ~ dnegbin(rate[i, s], phi[s])T(, upper_bound[s]); + rate[i, s] <- ifelse((phi[s] / (phi[s] + mus[i, s])) < min_eps, min_eps, + (phi[s] / (phi[s] + mus[i, s]))) } } @@ -397,14 +397,14 @@ add_base_dgam_lines = function(use_lv, stan = FALSE, offset = FALSE){ ## where the likelihood reduces to a 'base' model (Poisson) unless ## the data support overdispersion for (s in 1:n_series) { - r[s] <- 1 / r_inv[s] - r_inv[s] ~ dexp(5) + phi[s] <- 1 / phi_inv[s] + phi_inv[s] ~ dexp(5) } ## posterior predictions for (i in 1:n) { for (s in 1:n_series) { - ypred[i, s] ~ dnegbin(rate[i, s], r[s])T(, upper_bound[s]) + ypred[i, s] ~ dnegbin(rate[i, s], phi[s])T(, upper_bound[s]) } } diff --git a/R/add_poisson_lines.R b/R/add_poisson_lines.R index fabfaa09..cfafcea6 100644 --- a/R/add_poisson_lines.R +++ b/R/add_poisson_lines.R @@ -7,7 +7,7 @@ #' @return A modified `JAGS` model file add_poisson_lines = function(model_file, upper_bounds){ - odis_begin <- grep('r\\[s\\] <- ', model_file) - 4 + odis_begin <- grep('phi\\[s\\] <- ', model_file) - 4 odis_end <- odis_begin + 7 model_file <- model_file[-c(odis_begin:odis_end)] diff --git a/R/add_stan_data.R b/R/add_stan_data.R index 1d118a9e..510b0d04 100644 --- a/R/add_stan_data.R +++ b/R/add_stan_data.R @@ -35,16 +35,16 @@ add_stan_data = function(jags_file, stan_file, use_lv = FALSE, if(family == 'nb'){ stan_file[grep('// raw basis', stan_file) + 2] <- - '\n// negative binomial overdispersion\nvector[n_series] r_inv;\n' + '\n// negative binomial overdispersion\nvector[n_series] phi_inv;\n' stan_file[grep('// priors for smoothing', stan_file) + 2] <- paste0('\n// priors for overdispersion parameters\n', - 'r_inv ~ student_t(3, 0, 0.1);\n') + 'phi_inv ~ student_t(3, 0, 0.1);\n') to_negbin <- gsub('poisson_log', 'neg_binomial_2', stan_file[grep('y[i, s] ~ poisson', stan_file, fixed = T)]) stan_file[grep('y[i, s] ~ poisson', stan_file, fixed = T)] <- - gsub(');', ', inv(r_inv[s]));', to_negbin) + gsub(');', ', inv(phi_inv[s]));', to_negbin) add_exp_open <- gsub('\\(eta', '(exp(eta', stan_file[grep('y[i, s] ~ neg_binomial', stan_file, fixed = T)]) @@ -55,16 +55,16 @@ add_stan_data = function(jags_file, stan_file, use_lv = FALSE, stan_file[grep('matrix[n, n_series] ypred;', stan_file, fixed = T)] <- paste0('matrix[n, n_series] ypred;\n', - 'matrix[n, n_series] r_vec;\n', - 'vector[n_series] r;\n', - 'r = inv(r_inv);\n', + 'matrix[n, n_series] phi_vec;\n', + 'vector[n_series] phi;\n', + 'phi = inv(phi_inv);\n', 'for (s in 1:n_series) {\n', - 'r_vec[1:n,s] = rep_vector(r[s], n);\n}\n') + 'phi_vec[1:n,s] = rep_vector(phi[s], n);\n}\n') to_negbin <- gsub('poisson_log_rng', 'neg_binomial_2_rng', stan_file[grep('ypred[i, s] = poisson_log_rng', stan_file, fixed = T)]) stan_file[grep('ypred[i, s] = poisson_log_rng', stan_file, fixed = T)] <- - gsub(');', ', r_vec[i, s]);', to_negbin) + gsub(');', ', phi_vec[i, s]);', to_negbin) add_exp_open <- gsub('\\(eta', '(exp(eta', stan_file[grep('ypred[i, s] = neg_binomial', stan_file, fixed = T)]) diff --git a/R/add_trend_lines.R b/R/add_trend_lines.R index 04526231..61142914 100644 --- a/R/add_trend_lines.R +++ b/R/add_trend_lines.R @@ -59,7 +59,7 @@ add_trend_lines = function(model_file, stan = FALSE, 'real boundary;\n', 'int num_gp_basis;\n', 'num_gp_basis = min(20, n);\n', - 'matrix[n, num_gp_basis] gp_phi;\n\n', + 'matrix[n, num_gp_basis] gp_drift;\n\n', 'for (t in 1:n){\n', 'times[t] = t;\n', '}\n\n', @@ -67,7 +67,7 @@ add_trend_lines = function(model_file, stan = FALSE, 'times_cent = times - mean_times;\n', 'boundary = (5.0/4) * (max(times_cent) - min(times_cent));\n', 'for (m in 1:num_gp_basis){\n', - 'gp_phi[,m] = phi_SE(boundary, m, times_cent);\n', + 'gp_drift[,m] = drift_SE(boundary, m, times_cent);\n', '}\n') model_file[grep('##insert data', model_file)-1] <- @@ -77,7 +77,7 @@ add_trend_lines = function(model_file, stan = FALSE, 'lam = ((m*pi())/(2*L))^2;\n', 'return lam;\n', '}\n\n', - 'vector phi_SE(real L, int m, vector x) {\n', + 'vector drift_SE(real L, int m, vector x) {\n', 'vector[rows(x)] fi;\n', 'fi = 1/sqrt(L) * sin(m*pi()/(2*L) * (x+L));\n', 'return fi;\n', @@ -110,7 +110,7 @@ add_trend_lines = function(model_file, stan = FALSE, 'diag_SPD[m, s] = sqrt(spd_SE(0.25, rho_gp[s], sqrt(lambda_gp(boundary, m))));\n', '}\n}\n', 'SPD_beta = diag_SPD .* b_gp;\n', - 'LV_raw = gp_phi * SPD_beta;\n}\n') + 'LV_raw = gp_drift * SPD_beta;\n}\n') rho_line <- 'rho_gp ~ inv_gamma(1.499007, 5.670433);\n' @@ -133,8 +133,8 @@ add_trend_lines = function(model_file, stan = FALSE, model_file <- model_file[-c((grep('// derived latent trends', model_file)): (grep('// derived latent trends', model_file) + 5))] - model_file[grep('LV_raw = gp_phi * SPD_beta;', model_file, fixed = TRUE)] <- - paste0('LV_raw = gp_phi * SPD_beta;\n', + model_file[grep('LV_raw = gp_drift * SPD_beta;', model_file, fixed = TRUE)] <- + paste0('LV_raw = gp_drift * SPD_beta;\n', '// derived latent trends\n', 'for (i in 1:n){;\n', 'for (s in 1:n_series){\n', @@ -167,7 +167,7 @@ add_trend_lines = function(model_file, stan = FALSE, 'real boundary;\n', 'int num_gp_basis;\n', 'num_gp_basis = min(20, n);\n', - 'matrix[n, num_gp_basis] gp_phi;\n\n', + 'matrix[n, num_gp_basis] gp_drift;\n\n', 'for (t in 1:n){\n', 'times[t] = t;\n', '}\n\n', @@ -175,7 +175,7 @@ add_trend_lines = function(model_file, stan = FALSE, 'times_cent = times - mean_times;\n', 'boundary = (5.0/4) * (max(times_cent) - min(times_cent));\n', 'for (m in 1:num_gp_basis){\n', - 'gp_phi[,m] = phi_SE(boundary, m, times_cent);\n', + 'gp_drift[,m] = drift_SE(boundary, m, times_cent);\n', '}\n}\n\n', 'parameters {') @@ -186,7 +186,7 @@ add_trend_lines = function(model_file, stan = FALSE, 'lam = ((m*pi())/(2*L))^2;\n', 'return lam;\n', '}\n\n', - 'vector phi_SE(real L, int m, vector x) {\n', + 'vector drift_SE(real L, int m, vector x) {\n', 'vector[rows(x)] fi;\n', 'fi = 1/sqrt(L) * sin(m*pi()/(2*L) * (x+L));\n', 'return fi;\n', @@ -220,7 +220,7 @@ add_trend_lines = function(model_file, stan = FALSE, 'diag_SPD[m, s] = sqrt(spd_SE(alpha_gp[s], rho_gp[s], sqrt(lambda_gp(boundary, m))));\n', '}\n}\n', 'SPD_beta = diag_SPD .* b_gp;\n', - 'trend = gp_phi * SPD_beta;\n}\n') + 'trend = gp_drift * SPD_beta;\n}\n') rho_line <- 'rho_gp ~ inv_gamma(1.499007, 5.670433);\n' alpha_line <- 'alpha_gp ~ normal(0, 0.5);\n' @@ -293,26 +293,26 @@ add_trend_lines = function(model_file, stan = FALSE, if(drift){ if(use_lv){ model_file[grep('// raw basis', model_file) + 1] <- - paste0('row_vector[num_basis] b_raw;\n\n// latent factor drift terms\nvector[n_lv] phi;\n') + paste0('row_vector[num_basis] b_raw;\n\n// latent factor drift terms\nvector[n_lv] drift;\n') model_file[grep('LV_raw[1, j] ~ ', model_file, fixed = T)] <- - "LV_raw[1, j] ~ normal(phi[j], 0.1);" + "LV_raw[1, j] ~ normal(drift[j], 0.1);" model_file[grep('// dynamic factor estimates', model_file) + 6] <- - paste0('LV_raw[2:n, j] ~ normal(phi[j] + LV_raw[1:(n - 1), j], 0.1);') + paste0('LV_raw[2:n, j] ~ normal(drift[j] + LV_raw[1:(n - 1), j], 0.1);') } else { model_file[grep('// raw basis', model_file) + 1] <- - paste0('row_vector[num_basis] b_raw;\n\n// latent trend drift terms\nvector[n_series] phi;\n') + paste0('row_vector[num_basis] b_raw;\n\n// latent trend drift terms\nvector[n_series] drift;\n') model_file[grep('trend[1, s] ~ ', model_file, fixed = T)] <- - "trend[1, s] ~ normal(phi[s], sigma[s]);" + "trend[1, s] ~ normal(drift[s], sigma[s]);" model_file[grep('// trend estimates', model_file) + 6] <- - paste0('trend[2:n, s] ~ normal(phi[s] + trend[1:(n - 1), s], sigma[s]);') + paste0('trend[2:n, s] ~ normal(drift[s] + trend[1:(n - 1), s], sigma[s]);') } model_file[grep('model \\{', model_file) + 2] <- - paste0('\n// priors for trend parameters\nphi ~ normal(0, 0.1);\n') + paste0('\n// priors for trend parameters\ndrift ~ normal(0, 0.1);\n') model_file <- readLines(textConnection(model_file), n = -1) @@ -325,29 +325,29 @@ add_trend_lines = function(model_file, stan = FALSE, if(use_lv){ model_file[grep('// raw basis', model_file) + 1] <- paste0(c('row_vector[num_basis] b_raw;\n\n// latent factor AR1 terms\nvector[n_lv] ar1;\n\n'), - '// latent factor drift terms\nvector[n_lv] phi;') + '// latent factor drift terms\nvector[n_lv] drift;') model_file[grep('LV_raw[1, j] ~ ', model_file, fixed = T)] <- - "LV_raw[1, j] ~ normal(phi[j], 0.1);" + "LV_raw[1, j] ~ normal(drift[j], 0.1);" model_file[grep('// dynamic factor estimates', model_file) + 6] <- - paste0('LV_raw[2:n, j] ~ normal(phi[j] + ar1[j] * LV_raw[1:(n - 1), j], 0.1);') + paste0('LV_raw[2:n, j] ~ normal(drift[j] + ar1[j] * LV_raw[1:(n - 1), j], 0.1);') } else { model_file[grep('// raw basis', model_file) + 1] <- paste0(c('row_vector[num_basis] b_raw;\n\n// latent trend AR1 terms\nvector[n_series] ar1;\n\n'), - '// latent trend drift terms\nvector[n_series] phi;') + '// latent trend drift terms\nvector[n_series] drift;') model_file[grep('trend[1, s] ~ ', model_file, fixed = T)] <- - "trend[1, s] ~ normal(phi[s], sigma[s]);" + "trend[1, s] ~ normal(drift[s], sigma[s]);" model_file[grep('// trend estimates', model_file) + 6] <- - paste0('trend[2:n, s] ~ normal(phi[s] + ar1[s] * trend[1:(n - 1), s], sigma[s]);') + paste0('trend[2:n, s] ~ normal(drift[s] + ar1[s] * trend[1:(n - 1), s], sigma[s]);') } model_file[grep('model \\{', model_file) + 2] <- - paste0('\n// priors for AR parameters\nar1 ~ std_normal();\nphi ~ std_normal();\n') + paste0('\n// priors for AR parameters\nar1 ~ std_normal();\ndrift ~ std_normal();\n') } else { if(use_lv){ @@ -382,43 +382,43 @@ add_trend_lines = function(model_file, stan = FALSE, model_file[grep('// raw basis', model_file) + 1] <- paste0('row_vector[num_basis] b_raw;\n\n// latent factor AR1 terms\nvector[n_lv] ar1;\n\n', '// latent factor AR2 terms\nvector[n_lv] ar2;\n\n', - '// latent factor drift terms\nvector[n_lv] phi;') + '// latent factor drift terms\nvector[n_lv] drift;') model_file[grep('LV_raw[1, j] ~ ', model_file, fixed = T)] <- - "LV_raw[1, j] ~ normal(phi[j], 0.1);" + "LV_raw[1, j] ~ normal(drift[j], 0.1);" model_file <- model_file[-(grep('// dynamic factor estimates', model_file) + 5:7)] model_file[grep('// dynamic factor estimates', model_file) + 5] <- paste0('for (j in 1:n_lv) {\n', - 'LV_raw[2, j] ~ normal(phi[j] + LV_raw[1, j] * ar1[j], 0.1);\n', + 'LV_raw[2, j] ~ normal(drift[j] + LV_raw[1, j] * ar1[j], 0.1);\n', '}\n\n', 'for (i in 3:n) {\n', 'for (j in 1:n_lv) {\n', - 'LV_raw[i, j] ~ normal(phi[j] + ar1[j] * LV_raw[i - 1, j] + ar2[j] * LV_raw[i - 2, j], 0.1);\n', + 'LV_raw[i, j] ~ normal(drift[j] + ar1[j] * LV_raw[i - 1, j] + ar2[j] * LV_raw[i - 2, j], 0.1);\n', '}\n}\n') } else { model_file[grep('// raw basis', model_file) + 1] <- paste0('row_vector[num_basis] b_raw;\n\n// latent trend AR1 terms\nvector[n_series] ar1;\n\n', '// latent trend AR2 terms\nvector[n_series] ar2;\n\n', - '// latent trend drift terms\nvector[n_series] phi;') + '// latent trend drift terms\nvector[n_series] drift;') model_file[grep('trend[1, s] ~ ', model_file, fixed = T)] <- - "trend[1, s] ~ normal(phi[s], sigma[s]);" + "trend[1, s] ~ normal(drift[s], sigma[s]);" model_file <- model_file[-(grep('// trend estimates', model_file) + 5:7)] model_file[grep('// trend estimates', model_file) + 5] <- paste0('for (s in 1:n_series) {\n', - 'trend[2, s] ~ normal(phi[s] + trend[1, s] * ar1[s], sigma[s]);\n', + 'trend[2, s] ~ normal(drift[s] + trend[1, s] * ar1[s], sigma[s]);\n', '}\n\n', 'for (i in 3:n) {\n', 'for (s in 1:n_series) {\n', - 'trend[i, s] ~ normal(phi[s] + ar1[s] * trend[i - 1, s] + ar2[s] * trend[i - 2, s], sigma[s]);\n', + 'trend[i, s] ~ normal(drift[s] + ar1[s] * trend[i - 1, s] + ar2[s] * trend[i - 2, s], sigma[s]);\n', '}\n}\n') } model_file[grep('model \\{', model_file) + 2] <- - paste0('\n// priors for AR parameters\nar1 ~ std_normal();\nar2 ~ std_normal();\nphi ~ std_normal();\n') + paste0('\n// priors for AR parameters\nar1 ~ std_normal();\nar2 ~ std_normal();\ndrift ~ std_normal();\n') } else { if(use_lv){ @@ -470,54 +470,54 @@ add_trend_lines = function(model_file, stan = FALSE, paste0('row_vector[num_basis] b_raw;\n\n// latent factor AR1 terms\nvector[n_lv] ar1;\n\n', '// latent factor AR2 terms\nvector[n_lv] ar2;\n\n', '// latent factor AR3 terms\nvector[n_lv] ar3;\n\n', - '// latent factor drift terms\nvector[n_lv] phi;') + '// latent factor drift terms\nvector[n_lv] drift;') model_file[grep('LV_raw[1, s] ~ ', model_file, fixed = T)] <- - "LV_raw[1, s] ~ normal(phi[s], 0.1);" + "LV_raw[1, s] ~ normal(drift[s], 0.1);" model_file <- model_file[-(grep('// dynamic factor estimates', model_file) + 5:7)] model_file[grep('// dynamic factor estimates', model_file) + 5] <- paste0('for (j in 1:n_lv) {\n', - 'LV_raw[2, j] ~ normal(phi[j] + LV_raw[1, j] * ar1[j], 0.1);\n', + 'LV_raw[2, j] ~ normal(drift[j] + LV_raw[1, j] * ar1[j], 0.1);\n', '}\n\n', 'for (j in 1:n_lv) {\n', - 'LV_raw[3, j] ~ normal(phi[j] + LV_raw[2, j] * ar1[j] + LV_raw[1, j] * ar2[j], 0.1);\n', + 'LV_raw[3, j] ~ normal(drift[j] + LV_raw[2, j] * ar1[j] + LV_raw[1, j] * ar2[j], 0.1);\n', '}\n\n', 'for (i in 4:n) {\n', 'for (j in 1:n_lv) {\n', - 'LV_raw[i, j] ~ normal(phi[j] + ar1[j] * LV_raw[i - 1, j] + ar2[j] * LV_raw[i - 2, j] + ar3[j] * LV_raw[i - 3, j], 0.1);\n', + 'LV_raw[i, j] ~ normal(drift[j] + ar1[j] * LV_raw[i - 1, j] + ar2[j] * LV_raw[i - 2, j] + ar3[j] * LV_raw[i - 3, j], 0.1);\n', '}\n}\n') } else { model_file[grep('// raw basis', model_file) + 1] <- paste0('row_vector[num_basis] b_raw;\n\n// latent trend AR1 terms\nvector[n_series] ar1;\n\n', '// latent trend AR2 terms\nvector[n_series] ar2;\n\n', '// latent trend AR3 terms\nvector[n_series] ar3;\n\n', - '// latent trend drift terms\nvector[n_series] phi;') + '// latent trend drift terms\nvector[n_series] drift;') model_file[grep('trend[1, s] ~ ', model_file, fixed = T)] <- - "trend[1, s] ~ normal(phi[s], sigma[s]);" + "trend[1, s] ~ normal(drift[s], sigma[s]);" model_file <- model_file[-(grep('// trend estimates', model_file) + 5:7)] model_file[grep('// trend estimates', model_file) + 5] <- paste0('for (s in 1:n_series) {\n', - 'trend[2, s] ~ normal(phi[s] + trend[1, s] * ar1[s], sigma[s]);\n', + 'trend[2, s] ~ normal(drift[s] + trend[1, s] * ar1[s], sigma[s]);\n', '}\n\n', 'for (s in 1:n_series) {\n', - 'trend[3, s] ~ normal(phi[s] + trend[2, s] * ar1[s] + trend[1, s] * ar2[s], sigma[s]);\n', + 'trend[3, s] ~ normal(drift[s] + trend[2, s] * ar1[s] + trend[1, s] * ar2[s], sigma[s]);\n', '}\n\n', 'for (i in 4:n) {\n', 'for (s in 1:n_series) {\n', - 'trend[i, s] ~ normal(phi[s] + ar1[s] * trend[i - 1, s] + ar2[s] * trend[i - 2, s] + ar3[s] * trend[i - 3, s], sigma[s]);\n', + 'trend[i, s] ~ normal(drift[s] + ar1[s] * trend[i - 1, s] + ar2[s] * trend[i - 2, s] + ar3[s] * trend[i - 3, s], sigma[s]);\n', '}\n}\n') } model_file[grep('model \\{', model_file) + 2] <- paste0('\n// priors for AR parameters\nar1 ~ std_normal();\nar2 ~ std_normal();\nar3 ~ std_normal();\n', - 'phi ~ std_normal();\n') + 'drift ~ std_normal();\n') } else { @@ -589,11 +589,11 @@ add_trend_lines = function(model_file, stan = FALSE, if(drift){ model_file[grep('## latent factors evolve', model_file) + 5] <- - '\nfor (i in 2:n) {\nfor (j in 1:n_lv){\nLV_raw[i, j] ~ dnorm(phi[j] + LV_raw[i - 1, j], penalty[j])\n}\n}\n' + '\nfor (i in 2:n) {\nfor (j in 1:n_lv){\nLV_raw[i, j] ~ dnorm(drift[j] + LV_raw[i - 1, j], penalty[j])\n}\n}\n' } else { model_file[grep('## latent factors evolve', model_file) + 5] <- '\nfor (i in 2:n) {\nfor (j in 1:n_lv){\nLV_raw[i, j] ~ dnorm(LV_raw[i - 1, j], penalty[j])\n}\n}\n' - model_file <- model_file[-grep('phi\\[s\\] ~', model_file)] + model_file <- model_file[-grep('drift\\[s\\] ~', model_file)] } model_file <- readLines(textConnection(model_file), n = -1) @@ -614,7 +614,7 @@ add_trend_lines = function(model_file, stan = FALSE, } else { model_file[grep('## latent factors evolve', model_file) + 9] <- 'LV_raw[i, j] ~ dnorm(ar1[j]*LV_raw[i - 1, j], penalty[j])\n}' - model_file <- model_file[-grep('phi\\[s\\] ~', model_file)] + model_file <- model_file[-grep('drift\\[s\\] ~', model_file)] } model_file <- readLines(textConnection(model_file), n = -1) @@ -638,7 +638,7 @@ add_trend_lines = function(model_file, stan = FALSE, 'LV_raw[i, j] ~ dnorm(ar1[j]*LV_raw[i - 1, j] +' model_file[grep('## latent factors evolve', model_file) + 14] <- 'ar2[j]*LV_raw[i - 2, j], penalty[j])' - model_file <- model_file[-grep('phi\\[s\\] ~', model_file)] + model_file <- model_file[-grep('drift\\[s\\] ~', model_file)] } model_file <- readLines(textConnection(model_file), n = -1) model_file <- model_file[-grep('ar3\\[s\\] ~', model_file)] @@ -655,7 +655,7 @@ add_trend_lines = function(model_file, stan = FALSE, 'LV_raw[3, j] ~ dnorm(ar1[j]*LV_raw[2, j] + ar2[j]*LV_raw[1, j], penalty[j])' model_file[grep('## latent factors evolve', model_file) + 16] <- 'LV_raw[i, j] ~ dnorm(ar1[j]*LV_raw[i - 1, j] +' - model_file <- model_file[-grep('phi\\[s\\] ~', model_file)] + model_file <- model_file[-grep('drift\\[s\\] ~', model_file)] } model_file <- readLines(textConnection(model_file), n = -1) } @@ -669,14 +669,14 @@ add_trend_lines = function(model_file, stan = FALSE, if(drift){ model_file[grep('## trend estimates', model_file) + 2] <- - "trend[1, s] ~ dnorm(phi[s], tau[s])" + "trend[1, s] ~ dnorm(drift[s], tau[s])" model_file[grep('## trend estimates', model_file) + 4] <- - '\nfor (i in 2:n) {\nfor (s in 1:n_series){\ntrend[i, s] ~ dnorm(phi[s] + trend[i - 1, s], tau[s])\n}\n}\n' + '\nfor (i in 2:n) {\nfor (s in 1:n_series){\ntrend[i, s] ~ dnorm(drift[s] + trend[i - 1, s], tau[s])\n}\n}\n' } else { model_file[grep('## trend estimates', model_file) + 4] <- '\nfor (i in 2:n) {\nfor (s in 1:n_series){\ntrend[i, s] ~ dnorm(trend[i - 1, s], tau[s])\n}\n}\n' - model_file <- model_file[-grep('phi\\[s\\] ~', model_file)] + model_file <- model_file[-grep('drift\\[s\\] ~', model_file)] } model_file <- readLines(textConnection(model_file), n = -1) @@ -691,14 +691,14 @@ add_trend_lines = function(model_file, stan = FALSE, if(drift){ model_file[grep('## trend estimates', model_file) + 2] <- - "trend[1, s] ~ dnorm(phi[s], tau[s])" + "trend[1, s] ~ dnorm(drift[s], tau[s])" model_file[grep('## trend estimates', model_file) + 4] <- - '\nfor (i in 2:n) {\nfor (s in 1:n_series){\ntrend[i, s] ~ dnorm(phi[s] + ar1[s]*trend[i - 1, s], tau[s])\n}\n}\n' + '\nfor (i in 2:n) {\nfor (s in 1:n_series){\ntrend[i, s] ~ dnorm(drift[s] + ar1[s]*trend[i - 1, s], tau[s])\n}\n}\n' } else { model_file[grep('## trend estimates', model_file) + 4] <- '\nfor (i in 2:n) {\nfor (s in 1:n_series){\ntrend[i, s] ~ dnorm(ar1[s]*trend[i - 1, s], tau[s])\n}\n}\n' - model_file <- model_file[-grep('phi\\[s\\] ~', model_file)] + model_file <- model_file[-grep('drift\\[s\\] ~', model_file)] } model_file <- readLines(textConnection(model_file), n = -1) @@ -712,16 +712,16 @@ add_trend_lines = function(model_file, stan = FALSE, if(drift){ model_file[grep('## trend estimates', model_file) + 2] <- - "trend[1, s] ~ dnorm(phi[s], tau[s])" + "trend[1, s] ~ dnorm(drift[s], tau[s])" model_file[grep('## trend estimates', model_file) + 9] <- - '\nfor (i in 3:n) {\nfor (s in 1:n_series){\ntrend[i, s] ~ dnorm(phi[s] + ar1[s]*trend[i - 1, s] + ar2[s]*trend[i - 2, s], tau[s])\n}\n}\n' + '\nfor (i in 3:n) {\nfor (s in 1:n_series){\ntrend[i, s] ~ dnorm(drift[s] + ar1[s]*trend[i - 1, s] + ar2[s]*trend[i - 2, s], tau[s])\n}\n}\n' } else { model_file[grep('## trend estimates', model_file) + 6] <- 'trend[2, s] ~ dnorm(ar1[s]*trend[1, s], tau[s])' model_file[grep('## trend estimates', model_file) + 9] <- '\nfor (i in 3:n) {\nfor (s in 1:n_series){\ntrend[i, s] ~ dnorm(ar1[s]*trend[i - 1, s] + ar2[s]*trend[i - 2, s], tau[s])\n}\n}\n' - model_file <- model_file[-grep('phi\\[s\\] ~', model_file)] + model_file <- model_file[-grep('drift\\[s\\] ~', model_file)] } model_file <- readLines(textConnection(model_file), n = -1) model_file <- model_file[-grep('ar3\\[s\\] ~', model_file)] @@ -731,7 +731,7 @@ add_trend_lines = function(model_file, stan = FALSE, if(drift){ model_file[grep('## trend estimates', model_file) + 2] <- - "trend[1, s] ~ dnorm(phi[s], tau[s])" + "trend[1, s] ~ dnorm(drift[s], tau[s])" } else { model_file[grep('## trend estimates', model_file) + 6] <- @@ -740,7 +740,7 @@ add_trend_lines = function(model_file, stan = FALSE, 'trend[3, s] ~ dnorm(ar1[s]*trend[2, s] + ar2[s]*trend[1, s], tau[s])' model_file[grep('## trend estimates', model_file) + 15] <- 'trend[i, s] ~ dnorm(ar1[s]*trend[i - 1, s] + ar2[s]*trend[i - 2, s] + ar3[s]*trend[i - 3, s], tau[s])' - model_file <- model_file[-grep('phi\\[s\\] ~', model_file)] + model_file <- model_file[-grep('drift\\[s\\] ~', model_file)] } model_file <- readLines(textConnection(model_file), n = -1) } diff --git a/R/add_tweedie_lines.R b/R/add_tweedie_lines.R index 0456bb66..cecaaad5 100644 --- a/R/add_tweedie_lines.R +++ b/R/add_tweedie_lines.R @@ -10,7 +10,7 @@ add_tweedie_lines = function(model_file, upper_bounds){ rate_end <- rate_begin + 1 model_file <- model_file[-c(rate_begin:rate_end)] - odis_begin <- grep('r\\[s\\] <- ', model_file) - 4 + odis_begin <- grep('phi\\[s\\] <- ', model_file) - 4 odis_end <- odis_begin + 7 model_file <- model_file[-c(odis_begin:odis_end)] @@ -27,7 +27,7 @@ add_tweedie_lines = function(model_file, upper_bounds){ model_file[grep('linpred\\[i, s\\] ~', model_file)] <- ' linpred[i, s] ~ dgamma(shape[i, s, y_ind[i, s]], rate[i, s])\n twlambda[i, s] <-' model_file <- readLines(textConnection(model_file), n = -1) - model_file[grep('twlambda\\[i, s\\] <-', model_file)] <- ' twlambda[i, s] <- pow(mus[i, s], 2 - p) / (twdis[s] * (2 - p))\n N_pois[i, s] ~' + model_file[grep('twlambda\\[i, s\\] <-', model_file)] <- ' twlambda[i, s] <- pow(mus[i, s], 2 - p) / (phi[s] * (2 - p))\n N_pois[i, s] ~' model_file <- readLines(textConnection(model_file), n = -1) model_file[grep('N_pois\\[i, s\\] ~', model_file)] <- ' N_pois[i, s] ~ dpois(twlambda[i, s])T(1,)\n shape[i, s, 1] <-' @@ -39,7 +39,7 @@ add_tweedie_lines = function(model_file, upper_bounds){ model_file[grep('shape\\[i, s, 2\\] <-', model_file)] <- ' shape[i, s, 2] <- 1\n rate[i, s] <-' model_file <- readLines(textConnection(model_file), n = -1) - model_file[grep('rate\\[i, s\\] <-', model_file)] <- ' rate[i, s] <- 1 / (twdis[s] * (p - 1) * pow(mus[i, s], p - 1))\n pois_draw[i, s] ~' + model_file[grep('rate\\[i, s\\] <-', model_file)] <- ' rate[i, s] <- 1 / (phi[s] * (p - 1) * pow(mus[i, s], p - 1))\n pois_draw[i, s] ~' model_file <- readLines(textConnection(model_file), n = -1) model_file[grep('pois_draw\\[i, s\\] ~', model_file)] <- ' pois_draw[i, s] ~ dpois(mus[i, s])\n is_zero[i, s] <-' @@ -53,7 +53,7 @@ add_tweedie_lines = function(model_file, upper_bounds){ yind_begin <- grep('y_ind\\[i, s\\] <-', model_file) prior_line <- yind_begin + 2 - model_file[prior_line] <- '}\n\n## Tweedie power and overdispersion parameters\np <- 1.5\nfor (s in 1:n_series) {\n twdis_raw[s] ~ dnorm(0, 2)T(-3.5, 3.5);\n twdis[s] <- exp(twdis_raw[s])\n}' + model_file[prior_line] <- '}\n\n## Tweedie power and overdispersion parameters\np <- 1.5\nfor (s in 1:n_series) {\n phi_raw[s] ~ dnorm(0, 2)T(-3.5, 3.5);\n phi[s] <- exp(phi_raw[s])\n}' model_file <- readLines(textConnection(model_file), n = -1) return(model_file) diff --git a/R/eval_mvgam.R b/R/eval_mvgam.R index fab89f92..c4bf62b7 100644 --- a/R/eval_mvgam.R +++ b/R/eval_mvgam.R @@ -136,7 +136,13 @@ eval_mvgam = function(object, lvs <- lapply(seq_len(n_lv), function(lv){ inds_lv <- seq(lv, dim(MCMCvis::MCMCchains(object$model_output, 'LV'))[2], by = n_lv) lv_estimates <- MCMCvis::MCMCchains(object$model_output, 'LV')[,inds_lv] - lv_estimates[,(eval_timepoint - 2):eval_timepoint] + + if(trend_model == 'GP'){ + lv_estimates + } else { + lv_estimates[,(eval_timepoint - 2):eval_timepoint] + } + }) } else { ends <- seq(0, dim(MCMCvis::MCMCchains(object$model_output, 'LV'))[2], @@ -225,14 +231,14 @@ eval_mvgam = function(object, # Beta coefficients for GAM component betas <- MCMCvis::MCMCchains(object$model_output, 'b') - # Phi estimates for latent trend drift terms + # drift estimates for latent trend drift terms if(object$drift){ - phis <- MCMCvis::MCMCchains(object$model_output, 'phi') + drifts <- MCMCvis::MCMCchains(object$model_output, 'drift') } else { if(object$use_lv){ - phis <- matrix(0, nrow = NROW(betas), ncol = object$n_lv) + drifts <- matrix(0, nrow = NROW(betas), ncol = object$n_lv) } else { - phis <- matrix(0, nrow = NROW(betas), ncol = n_series) + drifts <- matrix(0, nrow = NROW(betas), ncol = n_series) } } @@ -295,26 +301,15 @@ eval_mvgam = function(object, } # Family-specific parameters - if(object$family == 'Negative Binomial'){ - sizes <- MCMCvis::MCMCchains(object$model_output, 'r') - } else { - sizes <- NULL - } - - if(object$family == 'Tweedie'){ - twdis <- MCMCvis::MCMCchains(object$model_output, 'twdis') - p <- matrix(1.5, nrow = NROW(betas), ncol = n_series) - } else { - twdis <- p <- NULL - } - family <- object$family + pars <- mvgam:::extract_family_pars(family = family, + object = object) # Generate sample sequence for n_samples - if(n_samples < dim(phis)[1]){ - sample_seq <- sample(seq_len(dim(phis)[1]), size = n_samples, replace = F) + if(n_samples < dim(drifts)[1]){ + sample_seq <- sample(seq_len(dim(drifts)[1]), size = n_samples, replace = F) } else { - sample_seq <- sample(seq_len(dim(phis)[1]), size = n_samples, replace = T) + sample_seq <- sample(seq_len(dim(drifts)[1]), size = n_samples, replace = T) } @@ -336,7 +331,7 @@ eval_mvgam = function(object, 'Xp', 'betas', 'taus', - 'phis', + 'drifts', 'ar1s', 'ar2s', 'ar3s', @@ -346,9 +341,7 @@ eval_mvgam = function(object, 'lvs', 'lv_coefs', 'trends', - 'sizes', - 'p', - 'twdis', + 'pars', 'n_series', 'upper_bounds'), envir = environment()) @@ -359,7 +352,7 @@ eval_mvgam = function(object, samp_index <- x # Sample drift and AR parameters - phi <- phis[samp_index, ] + drift <- drifts[samp_index, ] ar1 <- ar1s[samp_index, ] ar2 <- ar2s[samp_index, ] ar3 <- ar3s[samp_index, ] @@ -373,11 +366,6 @@ eval_mvgam = function(object, # Sample beta coefs betas <- betas[samp_index, ] - # Family-specific parameters - size <- sizes[samp_index, ] - twdis <- twdis[samp_index, ] - p <- p[samp_index] - if(use_lv){ # Sample a last state estimate for the latent variables last_lvs <- lapply(seq_along(lvs), function(lv){ @@ -390,15 +378,24 @@ eval_mvgam = function(object, })) # Run the latent variables forward fc_horizon timesteps - lv_preds <- do.call(rbind, lapply(seq_len(n_lv), function(lv){ - sim_ar3(phi = phi[lv], - ar1 = ar1[lv], - ar2 = ar2[lv], - ar3 = ar3[lv], - tau = tau[lv], - last_trends = last_lvs[[lv]], - h = fc_horizon) - })) + if(trend_model == 'GP'){ + lv_preds <- do.call(rbind, lapply(seq_len(n_lv), function(lv){ + sim_gp(alpha_gp = alpha_gp[lv], + rho_gp = rho_gp[lv], + last_trends = last_lvs[[lv]], + h = fc_horizon) + })) + } else { + lv_preds <- do.call(rbind, lapply(seq_len(n_lv), function(lv){ + sim_ar3(drift = drift[lv], + ar1 = ar1[lv], + ar2 = ar2[lv], + ar3 = ar3[lv], + tau = tau[lv], + last_trends = last_lvs[[lv]], + h = fc_horizon) + })) + } series_fcs <- lapply(seq_len(n_series), function(series){ trend_preds <- as.numeric(t(lv_preds) %*% lv_coefs[series,]) @@ -406,13 +403,21 @@ eval_mvgam = function(object, Xpmat <- cbind(Xp[which(as.numeric(data_assim$series) == series),], trend_preds) attr(Xpmat, 'model.offset') <- attr(Xp, 'model.offset') - mvgam_predict(family = family, - Xp = Xpmat, - type = 'response', - betas = c(betas, 1), - size = size[series], - p = p, - twdis = twdis[series]) + # Family-specific parameters + par_extracts <- lapply(seq_along(pars), function(x){ + if(is.matrix(pars[[x]])){ + pars[[x]][samp_index, series] + } else { + pars[[x]][samp_index] + } + }) + names(par_extracts) <- names(pars) + + mvgam:::mvgam_predict(family = family, + Xp = Xpmat, + type = 'response', + betas = c(betas, 1), + family_pars = par_extracts) }) } else { @@ -429,7 +434,7 @@ eval_mvgam = function(object, last_trends = last_trends[[series]], h = fc_horizon) } else { - trend_preds <- sim_ar3(phi = phi[series], + trend_preds <- sim_ar3(drift = drift[series], ar1 = ar1[series], ar2 = ar2[series], ar3 = ar3[series], @@ -441,13 +446,21 @@ eval_mvgam = function(object, Xpmat <- cbind(Xp[which(as.numeric(data_assim$series) == series),], trend_preds) attr(Xpmat, 'model.offset') <- attr(Xp, 'model.offset') - mvgam_predict(family = family, + # Family-specific parameters + par_extracts <- lapply(seq_along(pars), function(x){ + if(is.matrix(pars[[x]])){ + pars[[x]][samp_index, series] + } else { + pars[[x]][samp_index] + } + }) + names(par_extracts) <- names(pars) + + mvgam:::mvgam_predict(family = family, Xp = Xpmat, type = 'response', betas = c(betas, 1), - size = size[series], - p = p, - twdis = twdis[series]) + family_pars = par_extracts) }) } @@ -462,7 +475,7 @@ eval_mvgam = function(object, samp_index <- x # Sample drift and AR parameters - phi <- phis[samp_index, ] + drift <- drifts[samp_index, ] ar1 <- ar1s[samp_index, ] ar2 <- ar2s[samp_index, ] ar3 <- ar3s[samp_index, ] @@ -476,11 +489,6 @@ eval_mvgam = function(object, # Sample beta coefs betas <- betas[samp_index, ] - # Family-specific parameters - size <- sizes[samp_index, ] - twdis <- twdis[samp_index, ] - p <- p[samp_index] - if(use_lv){ last_lvs <- lapply(seq_along(lvs), function(lv){ lvs[[lv]][samp_index, ] @@ -493,7 +501,7 @@ eval_mvgam = function(object, # Run the latent variables forward fc_horizon timesteps lv_preds <- do.call(rbind, lapply(seq_len(n_lv), function(lv){ - sim_ar3(phi = phi[lv], + sim_ar3(drift = drift[lv], ar1 = ar1[lv], ar2 = ar2[lv], ar3 = ar3[lv], @@ -507,13 +515,21 @@ eval_mvgam = function(object, Xpmat <- cbind(Xp[which(as.numeric(data_assim$series) == series),], trend_preds) attr(Xpmat, 'model.offset') <- attr(Xp, 'model.offset') - mvgam_predict(family = family, - Xp = Xpmat, - type = 'response', - betas = c(betas, 1), - size = size[series], - p = p, - twdis = twdis[series]) + # Family-specific parameters + par_extracts <- lapply(seq_along(pars), function(x){ + if(is.matrix(pars[[x]])){ + pars[[x]][samp_index, series] + } else { + pars[[x]][samp_index] + } + }) + names(par_extracts) <- names(pars) + + mvgam:::mvgam_predict(family = family, + Xp = Xpmat, + type = 'response', + betas = c(betas, 1), + family_pars = par_extracts) }) } else { @@ -530,7 +546,7 @@ eval_mvgam = function(object, last_trends = last_trends[[series]], h = fc_horizon) } else { - trend_preds <- sim_ar3(phi = phi[series], + trend_preds <- sim_ar3(drift = drift[series], ar1 = ar1[series], ar2 = ar2[series], ar3 = ar3[series], @@ -542,13 +558,21 @@ eval_mvgam = function(object, Xpmat <- cbind(Xp[which(as.numeric(data_assim$series) == series),], trend_preds) attr(Xpmat, 'model.offset') <- attr(Xp, 'model.offset') - mvgam_predict(family = family, - Xp = Xpmat, - type = 'response', - betas = c(betas, 1), - size = size[series], - p = p, - twdis = twdis[series]) + # Family-specific parameters + par_extracts <- lapply(seq_along(pars), function(x){ + if(is.matrix(pars[[x]])){ + pars[[x]][samp_index, series] + } else { + pars[[x]][samp_index] + } + }) + names(par_extracts) <- names(pars) + + mvgam:::mvgam_predict(family = family, + Xp = Xpmat, + type = 'response', + betas = c(betas, 1), + family_pars = par_extracts) }) } diff --git a/R/families.R b/R/families.R new file mode 100644 index 00000000..9063db5a --- /dev/null +++ b/R/families.R @@ -0,0 +1,123 @@ +#' Generic family information +#' @param family \code{character} +#' @param object Fitted model of class `mvgam` +#' @noRd +#' +#' +family_to_fullname = function(family){ + if(family == 'nb'){ + 'Negative Binomial' + } else if(family == 'poisson'){ + 'Poisson' + } else if(family == 'tw'){ + 'Tweedie' + } else if(family == 'beta'){ + 'Beta' + } else if(family == 'normal'){ + 'Normal' + } else if(family == 'lognormal'){ + 'LogNormal' + } else if(family == 'student'){ + 'Student-T' + } else if(family == 'gamma'){ + 'Gamma' + } +} + +# This function converts the fullnames to an appropriate family for +# fitting a reduced mgcv model to create the GAM structure and to allow +# for easier prediction post-fitting; it is needed because mgcv cannot handle all of +fullname_to_family = function(family){ + if(family == 'Negative Binomial'){ + 'nb' + } else if(family == 'Poisson'){ + 'poisson' + } else if(family == 'Tweedie'){ + 'Tweedie(p=1.5)' + } else if(family == 'Beta'){ + 'betar' + } else if(family == 'Normal'){ + 'gaussian' + } else if(family == 'LogNormal'){ + 'Gamma' + } else if(family == 'Student-T'){ + 'gaussian' + } else if(family == 'Gamma'){ + 'Gamma' + } +} + + +family_par_names = function(family){ + + # Parameters to monitor / extract depending on the observation family + if(family %in% c('Normal', + 'LogNormal')){ + out <- c('sigma_obs') + } + + if(family == 'Student-T'){ + out <- c('sigma_obs', 'nu_obs') + } + + if(family == 'Gamma'){ + out <- c('shape') + } + + if(family %in% c('Beta', + 'Negative Binomial', + 'Tweedie')){ + out <- c('phi') + } + + if(family == 'Poisson'){ + out <- c() + } + + out +} + +family_links = function(family){ + if(family %in% c('Normal', + 'LogNormal', + 'Student-T')){ + out <- c('identity') + } + + if(family %in% c('Negative Binomial', + 'Tweedie', + 'Poisson', + 'Gamma')){ + out <- c('log') + } + + if(family == 'Beta'){ + out <- 'logit' + } + + out +} + +extract_family_pars = function(family, object){ + + # Get names of parameters to extract + pars_to_extract <- family_par_names(family) + + # Extract into a named list + if(length(pars_to_extract) > 0){ + out <- vector(mode = 'list') + for(i in 1:length(pars_to_extract)){ + out[[i]] <- MCMCvis::MCMCchains(object$model_output, + params = pars_to_extract[i]) + } + names(out) <- pars_to_extract + + } else { + out <- list() + } + + # Return list of extracted posterior parameter samples + out +} + + diff --git a/R/forecast.mvgam.R b/R/forecast.mvgam.R index 1fe277f3..c1522e98 100644 --- a/R/forecast.mvgam.R +++ b/R/forecast.mvgam.R @@ -190,21 +190,9 @@ forecast.mvgam = function(object, newdata, data_test, series = 1, # Family of model family <- object$family - # Negative binomial size estimate - if(family == 'Negative Binomial'){ - sizes <- MCMCvis::MCMCchains(object$model_output, 'r') - } else { - sizes <- NULL - } - - # Tweedie parameters - if(family == 'Tweedie'){ - twdiss <- MCMCvis::MCMCchains(object$model_output, 'twdis') - ps <- matrix(1.5, nrow = NROW(betas), ncol = NCOL(object$ytimes)) - } else { - twdiss <- NULL - ps <- NULL - } + # Family-specific parameters + pars <- mvgam:::extract_family_pars(family = family, + object = object) # Latent trend precisions and loadings if(object$use_lv){ @@ -228,7 +216,7 @@ forecast.mvgam = function(object, newdata, data_test, series = 1, }) } else { - if(object$trend_model %in% c('None')){ + if(object$trend_model %in% c('GP','None')){ taus <- NULL } else { taus <- MCMCvis::MCMCchains(object$model_output, 'tau') @@ -296,14 +284,14 @@ forecast.mvgam = function(object, newdata, data_test, series = 1, lvs <- NULL } - # Phi estimates for latent trend drift terms + # Estimates for latent trend drift terms if(object$drift){ - phis <- MCMCvis::MCMCchains(object$model_output, 'phi') + drifts <- MCMCvis::MCMCchains(object$model_output, 'drift') } else { if(object$use_lv){ - phis <- matrix(0, nrow = NROW(betas), ncol = object$n_lv) + drifts <- matrix(0, nrow = NROW(betas), ncol = object$n_lv) } else { - phis <- matrix(0, nrow = NROW(betas), ncol = NCOL(object$ytimes)) + drifts <- matrix(0, nrow = NROW(betas), ncol = NCOL(object$ytimes)) } } @@ -380,18 +368,17 @@ forecast.mvgam = function(object, newdata, data_test, series = 1, setDefaultCluster(cl) clusterExport(NULL, c('use_lv', 'lvs', - 'phis', + 'drifts', 'ar1s', 'ar2s', 'ar3s', 'taus', 'alpha_gps', 'rho_gps', + 'pars', 'lv_coefs', 'betas', - 'sizes', - 'twdiss', - 'ps', + 'series', 'series_test', 'trend_estimates', 'Xp'), @@ -411,7 +398,7 @@ forecast.mvgam = function(object, newdata, data_test, series = 1, if(is.null(alpha_gps)){ # Sample AR parameters - phi <- phis[samp_index, ] + drift <- drifts[samp_index, ] ar1 <- ar1s[samp_index, ] ar2 <- ar2s[samp_index, ] ar3 <- ar3s[samp_index, ] @@ -421,15 +408,20 @@ forecast.mvgam = function(object, newdata, data_test, series = 1, tau <- taus[samp_index,] } else { - phi <- ar1 <- ar2 <- ar3 <- tau <- 0 + drift <- ar1 <- ar2 <- ar3 <- tau <- 0 alpha_gp <- alpha_gps[samp_index, ] rho_gp <- rho_gps[samp_index, ] } # Family-specific parameters - size <- sizes[samp_index, ] - twdis <- twdiss[samp_index, ] - p <- ps[samp_index] + par_extracts <- lapply(seq_along(pars), function(x){ + if(is.matrix(pars[[x]])){ + pars[[x]][samp_index, series] + } else { + pars[[x]][samp_index] + } + }) + names(par_extracts) <- names(pars) if(use_lv){ # Sample a last state estimate for the latent variables @@ -442,7 +434,7 @@ forecast.mvgam = function(object, newdata, data_test, series = 1, if(is.null(alpha_gps)){ next_lvs <- do.call(cbind, lapply(seq_along(lvs), function(lv){ - sim_ar3(phi = phi[lv], + sim_ar3(drift = drift[lv], ar1 = ar1[lv], ar2 = ar2[lv], ar3 = ar3[lv], @@ -465,7 +457,7 @@ forecast.mvgam = function(object, newdata, data_test, series = 1, } else { # Run the trends forward if(is.null(alpha_gps)){ - trends <- sim_ar3(phi = phi[series], + trends <- sim_ar3(drift = drift[series], ar1 = ar1[series], ar2 = ar2[series], ar3 = ar3[series], @@ -491,9 +483,7 @@ forecast.mvgam = function(object, newdata, data_test, series = 1, Xp = Xpmat, type = type, betas = c(betas, 1), - size = size[series], - p = p, - twdis = twdis[series]) + family_pars = par_extracts) } out diff --git a/R/get_monitor_pars.R b/R/get_monitor_pars.R index 58517a8b..6e83bb09 100644 --- a/R/get_monitor_pars.R +++ b/R/get_monitor_pars.R @@ -2,7 +2,7 @@ #' #' #' @export -#' @param family \code{character}. Must be either 'nb' (for Negative Binomial), 'tw' (for Tweedie) or 'poisson' +#' @param family \code{character} #' @param smooths_included Logical. Are smooth terms included in the model formula? #' @param use_lv Logical (use latent variable trends or not) #' @param trend_model The type of trend model used @@ -11,19 +11,18 @@ get_monitor_pars = function(family, smooths_included = TRUE, use_lv, trend_model, drift){ + family <- match.arg(arg = family, choices = c("nb", "poisson", + "tw", "beta", + "normal", "lognormal", + "student", "gamma")) + if(smooths_included){ param <- c('rho', 'b', 'ypred', 'mus') } else { param <- c('b', 'ypred', 'mus') } - if(family == 'nb'){ - param <- c(param, 'r') - } - - if(family == 'tw'){ - param <- c(param, 'twdis') - } + param <- c(param, family_par_names(family_to_fullname(family))) if(use_lv){ if(trend_model == 'RW'){ @@ -77,7 +76,7 @@ get_monitor_pars = function(family, smooths_included = TRUE, if(trend_model != 'None'){ if(drift){ - param <- c(param, 'phi') + param <- c(param, 'drift') } } param diff --git a/R/get_mvgam_priors.R b/R/get_mvgam_priors.R index 75e18054..a692fe06 100644 --- a/R/get_mvgam_priors.R +++ b/R/get_mvgam_priors.R @@ -692,29 +692,29 @@ get_mvgam_priors = function(formula, drift_df <- NULL } - # Extract information for family-specific overdispsersion parameters + # Extract information for family-specific overdispersion parameters if(family == 'nb'){ if(use_stan){ - nb_df <- data.frame(param_name = c('r_inv'), + nb_df <- data.frame(param_name = c('phi_inv'), param_length = length(unique(data_train$series)), param_info = c('inverse of NB dispsersion'), - prior = c('r_inv ~ student_t(3, 0, 0.1);'), + prior = c('phi_inv ~ student_t(3, 0, 0.1);'), example_change = c( paste0( - 'r_inv ~ normal(', + 'phi_inv ~ normal(', round(runif(min = -1, max = 1, n = 1), 2), ', ', round(runif(min = 0.1, max = 1, n = 1), 2), ');' ))) } else { - nb_df <- data.frame(param_name = c('r_inv'), + nb_df <- data.frame(param_name = c('phi_inv'), param_length = length(unique(data_train$series)), param_info = c('inverse of NB dispsersion'), - prior = c('r_inv[s] ~ dexp(5)'), + prior = c('phi_inv[s] ~ dexp(5)'), example_change = c( paste0( - 'r_inv[s] ~ dnorm(', + 'phi_inv[s] ~ dnorm(', round(runif(min = -1, max = 1, n = 1), 2), ', ', round(runif(min = 0.1, max = 1, n = 1), 2), @@ -727,13 +727,13 @@ get_mvgam_priors = function(formula, } if(family == 'tw'){ - tw_df <- data.frame(param_name = c('twdis_raw'), + tw_df <- data.frame(param_name = c('phi_raw'), param_length = length(unique(data_train$series)), param_info = c('log of Tweedie dispsersion (for each series s)'), - prior = c('twdis_raw[s] ~ dnorm(0, 2)T(-3.5, 3.5)'), + prior = c('phi_raw[s] ~ dnorm(0, 2)T(-3.5, 3.5)'), example_change = c( paste0( - 'twdis_raw[s] ~ dnorm(', + 'phi_raw[s] ~ dnorm(', round(runif(min = -1, max = 1, n = 1), 2), ', ', round(runif(min = 0.5, max = 5, n = 1), 2), diff --git a/R/get_mvgam_resids.R b/R/get_mvgam_resids.R index a68fac13..975ba206 100644 --- a/R/get_mvgam_resids.R +++ b/R/get_mvgam_resids.R @@ -40,9 +40,9 @@ sample_seq <- 1:NROW(preds) draw_seq <- sample(sample_seq, length(sample_seq), replace = FALSE) if(family == 'Negative Binomial'){ - rs <- MCMCvis::MCMCchains(object$model_output, 'r') + phis <- MCMCvis::MCMCchains(object$model_output, 'phi') } else { - rs <- NULL + phis <- NULL } # Functions for calculating randomised quantile (Dunn-Smyth) residuals @@ -146,7 +146,7 @@ clusterExport(NULL, c('ds_resids_nb', 'obs_series', 'series_levels', 'family', - 'rs', + 'phis', 'preds', 'ends', 'starts', @@ -204,7 +204,7 @@ series_resids <- pbapply::pblapply(seq_len(n_series), function(series){ } if(family == 'Negative Binomial'){ - size <- rs[,series] + size <- phis[,series] size_mat <- matrix(rep(size, NCOL(preds)), ncol = NCOL(preds)) resids <- matrix(ds_resids_nb(truth = as.vector(truth_mat), diff --git a/R/mvgam.R b/R/mvgam.R index 3604f7ee..07eeaba8 100644 --- a/R/mvgam.R +++ b/R/mvgam.R @@ -347,10 +347,30 @@ mvgam = function(formula, data_test <- newdata } - if(class(family) == 'family'){ + # Process family argument + if(any(class(family) == 'family')){ family <- family$family + if(family == 'negative binomial'){ + family <- 'nb' + } + if(family == 'Beta regression'){ + family <- 'beta' + } + if(grepl('Tweedie', family)){ + family <- 'tw' + } + if(family == 'gaussian'){ + family <- 'normal' + } + if(family == 'Gamma'){ + family <- 'gamma' + } } - family <- match.arg(arg = family, choices = c("nb", "poisson", "tw")) + + family <- match.arg(arg = family, choices = c("nb", "poisson", + "tw", "beta", + "normal", "lognormal", + "student", "gamma")) if(sign(chains) != 1){ stop('argument "chains" must be a positive integer', @@ -511,82 +531,11 @@ mvgam = function(formula, drop.unused.levels = FALSE, maxit = 30) - # if(!missing(knots)){ - # - # # Initiate the GAM model using mgcv so that the linear predictor matrix can be easily calculated - # # when simulating from the Bayesian model later on; we don't need convergence, just enough - # # iterations to get a sensible set of priors for any parametric coefficients - # if(family == 'nb'){ - # ss_gam <- mgcv::gam(formula(formula), - # data = data_train, - # method = "REML", - # family = nb(), - # knots = knots, - # control = list(nthreads = min(4, parallel::detectCores()-1), - # maxit = 50), - # drop.unused.levels = FALSE) - # } else if(family == 'poisson'){ - # ss_gam <- mgcv::gam(formula(formula), - # data = data_train, - # method = "REML", - # family = poisson(), - # knots = knots, - # control = list(nthreads = min(4, parallel::detectCores()-1), - # maxit = 50), - # drop.unused.levels = FALSE) - # } else { - # ss_gam <- mgcv::gam(formula(formula), - # data = data_train, - # method = "REML", - # family = tw(), - # knots = knots, - # control = list(nthreads = min(4, parallel::detectCores()-1), - # maxit = 50), - # drop.unused.levels = FALSE) - # } - # - # } else { - # if(family == 'nb'){ - # ss_gam <- mgcv::gam(formula(formula), - # data = data_train, - # method = "REML", - # family = nb(), - # control = list(nthreads = min(4, parallel::detectCores()-1), - # maxit = 50), - # drop.unused.levels = FALSE) - # } else if(family == 'poisson'){ - # ss_gam <- mgcv::gam(formula(formula), - # data = data_train, - # method = "REML", - # family = poisson(), - # control = list(nthreads = min(4, parallel::detectCores()-1), - # maxit = 50), - # drop.unused.levels = FALSE) - # } else { - # ss_gam <- mgcv::gam(formula(formula), - # data = data_train, - # method = "REML", - # family = tw(), - # control = list(nthreads = min(4, parallel::detectCores()-1), - # maxit = 50), - # drop.unused.levels = FALSE) - # } - # - # } - # Fill in missing observations in data_train so the size of the dataset is correct when # building the JAGS model - suppressWarnings(data_train$y[which(is.na(data_train$y))] <- round(predict(gam(y ~ time, data = data_train, - method = "REML", - family = poisson(), - control = list(nthreads = parallel::detectCores()-1, - maxit = 50), - drop.unused.levels = FALSE), - newdata = data.frame(time = data_train$time), - type = 'response'), - 0)[which(is.na(data_train$y))]) - - # Make jags file and appropriate data structures; note this has to use Poisson but the + suppressWarnings(data_train$y[which(is.na(data_train$y))] <- 0) + + # Make JAGS file and appropriate data structures; note this has to use Poisson but the # resulting JAGS file will be modified to accommodate the specified response distribution accordingly if(length(ss_gam$smooth) == 0){ smooths_included <- FALSE @@ -672,7 +621,7 @@ mvgam = function(formula, lines_remove <- c(1:grep('## response', base_model)) base_model <- base_model[-lines_remove] - # Any parametric effects in the gam (particularly the intercept) need to get more sensible priors to ensure they + # Any parametric effects in the gam (particularly the intercept) need sensible priors to ensure they # do not directly compete with the latent trends if(any(grepl('Parametric effect priors', base_model))){ @@ -751,8 +700,10 @@ mvgam = function(formula, # Remove the fakery lines if they were added if(!smooths_included){ - base_model <- base_model[-c(grep('## prior for s(fakery)', trimws(base_model), fixed = TRUE): - (grep('## prior for s(fakery)', trimws(base_model), fixed = TRUE) + 7))] + base_model <- base_model[-c(grep('## prior for s(fakery)', + trimws(base_model), fixed = TRUE): + (grep('## prior for s(fakery)', + trimws(base_model), fixed = TRUE) + 7))] } # Add replacement lines for priors, trends and the linear predictor @@ -771,8 +722,8 @@ mvgam = function(formula, } else { if(missing(upper_bounds)){ - model_file[grep('y\\[i, s\\] ~', model_file)] <- ' y[i, s] ~ dnegbin(rate[i, s], r[s])' - model_file[grep('ypred\\[i, s\\] ~', model_file)] <- ' ypred[i, s] ~ dnegbin(rate[i, s], r[s])' + model_file[grep('y\\[i, s\\] ~', model_file)] <- ' y[i, s] ~ dnegbin(rate[i, s], phi[s])' + model_file[grep('ypred\\[i, s\\] ~', model_file)] <- ' ypred[i, s] ~ dnegbin(rate[i, s], phi[s])' } } @@ -819,7 +770,8 @@ mvgam = function(formula, if(inherits(lp_test, 'try-error')){ testdat <- data.frame(time = data_test$time) - terms_include <- names(ss_gam$coefficients)[which(!names(ss_gam$coefficients) %in% '(Intercept)')] + terms_include <- names(ss_gam$coefficients)[which(!names(ss_gam$coefficients) + %in% '(Intercept)')] if(length(terms_include) > 0){ newnames <- vector() newnames[1] <- 'time' @@ -1124,13 +1076,13 @@ mvgam = function(formula, if(smooths_included){ inits <- function() { list(b_raw = runif(model_data$num_basis, -2, 2), - r_inv = runif(NCOL(model_data$ytimes), 1, 50), + phi_inv = runif(NCOL(model_data$ytimes), 1, 50), lambda = runif(model_data$n_sp, 5, 25)) } } else { inits <- function() { list(b_raw = runif(model_data$num_basis, -2, 2), - r_inv = runif(NCOL(model_data$ytimes), 1, 50)) + phi_inv = runif(NCOL(model_data$ytimes), 1, 50)) } } @@ -1138,14 +1090,14 @@ mvgam = function(formula, if(smooths_included){ inits <- function() { list(b_raw = runif(model_data$num_basis, -2, 2), - r_inv = runif(NCOL(model_data$ytimes), 1, 50), + phi_inv = runif(NCOL(model_data$ytimes), 1, 50), sigma = runif(model_data$n_series, 0.075, 1), lambda = runif(model_data$n_sp, 5, 25)) } } else { inits <- function() { list(b_raw = runif(model_data$num_basis, -2, 2), - r_inv = runif(NCOL(model_data$ytimes), 1, 50), + phi_inv = runif(NCOL(model_data$ytimes), 1, 50), sigma = runif(model_data$n_series, 0.075, 1)) } } @@ -1315,13 +1267,13 @@ mvgam = function(formula, if(smooths_included){ inits <- function() { list(b_raw = runif(model_data$num_basis, -2, 2), - r_inv = runif(NCOL(model_data$ytimes), 1, 50), + phi_inv = runif(NCOL(model_data$ytimes), 1, 50), lambda = runif(model_data$n_sp, 5, 25)) } } else { inits <- function() { list(b_raw = runif(model_data$num_basis, -2, 2), - r_inv = runif(NCOL(model_data$ytimes), 1, 50)) + phi_inv = runif(NCOL(model_data$ytimes), 1, 50)) } } @@ -1329,14 +1281,14 @@ mvgam = function(formula, if(smooths_included){ inits <- function() { list(b_raw = runif(model_data$num_basis, -2, 2), - r_inv = runif(NCOL(model_data$ytimes), 1, 50), + phi_inv = runif(NCOL(model_data$ytimes), 1, 50), sigma = runif(model_data$n_series, 0.075, 1), lambda = runif(model_data$n_sp, 5, 25)) } } else { inits <- function() { list(b_raw = runif(model_data$num_basis, -2, 2), - r_inv = runif(NCOL(model_data$ytimes), 1, 50), + phi_inv = runif(NCOL(model_data$ytimes), 1, 50), sigma = runif(model_data$n_series, 0.075, 1)) } } diff --git a/R/mvgam_predict.R b/R/mvgam_predict.R index c90daa3a..0e05ab9e 100644 --- a/R/mvgam_predict.R +++ b/R/mvgam_predict.R @@ -5,20 +5,31 @@ #' @param type When this has the value \code{link} (default) the linear predictor is calculated on the log link scale. #' When \code{response} is used, the predictions take uncertainty in the observation process into account to return #' predictions on the outcome scale -#' @param ... Additional arguments for each specific observation process (i.e. +#' @param family_pars Additional arguments for each specific observation process (i.e. #' overdispersion parameter if `family == "nb"`) +#' @param density logical. Rather than calculating a prediction, evaluate the likelihood. +#' Use this option when particle filtering +#' @param truth Observation to use for evaluating the likelihood (if `density == TRUE`) #' @details A generic prediction function that will make it easier to add new #' response distributions in future #' @noRd #' #' -mvgam_predict = function(Xp, family, betas, type = 'link', ...){ +mvgam_predict = function(Xp, family, betas, + type = 'link', + family_pars, + density = FALSE, + truth = NULL){ # Poisson observations if(family == 'Poisson'){ if(type == 'link'){ out <- as.vector((matrix(Xp, ncol = NCOL(Xp)) %*% betas) + attr(Xp, 'model.offset')) + if(density){ + out <- dpois(truth, lambda = exp(out)) + } + } else { out <- rpois(n = NROW(Xp), lambda = exp(((matrix(Xp, ncol = NCOL(Xp)) %*% @@ -27,33 +38,50 @@ mvgam_predict = function(Xp, family, betas, type = 'link', ...){ } } - # Negative Binomial observations (requires argument 'size') + # Negative Binomial observations (requires argument 'phi') if(family == 'Negative Binomial'){ if(type == 'link'){ out <- as.vector((matrix(Xp, ncol = NCOL(Xp)) %*% betas) + attr(Xp, 'model.offset')) + + if(density){ + out <- dnbinom(truth, mu = exp(out), + size = family_pars$phi) + } + } else { out <- rnbinom(n = NROW(Xp), mu = exp(((matrix(Xp, ncol = NCOL(Xp)) %*% betas)) + attr(Xp, 'model.offset')), - size = list(...)$size) + size = family_pars$phi) } } - # Tweedie observations (requires arguments 'p' and 'twdis') + # Tweedie observations (requires argument 'phi') if(family == 'Tweedie'){ if(type == 'link'){ out <- as.vector((matrix(Xp, ncol = NCOL(Xp)) %*% betas) + attr(Xp, 'model.offset')) + + if(density){ + out <- exp(mgcv::ldTweedie(y = truth, + mu = exp(out), + # Power parameter is fixed + p = 1.5, + phi = family_pars$phi, + all.derivs = F)[,1]) + } + } else { out <- rpois(n = NROW(Xp), lambda = mgcv::rTweedie( mu = exp(((matrix(Xp, ncol = NCOL(Xp)) %*% betas)) + attr(Xp, 'model.offset')), - p = list(...)$p, - phi = list(...)$twdis)) + # Power parameter is fixed + p = 1.5, + phi = family_pars$phi)) } } @@ -90,7 +118,7 @@ sim_gp = function(last_trends, h, rho_gp, alpha_gp){ #' Generic AR3 simulation function #' @param last_trends Vector of trend estimates leading up to the current timepoint #' @param h \code{integer} specifying the forecast horizon -#' @param phi drift parameter +#' @param drift drift parameter #' @param ar1 AR1 parameter #' @param ar2 AR2 parameter #' @param ar3 AR3 parameter @@ -98,13 +126,13 @@ sim_gp = function(last_trends, h, rho_gp, alpha_gp){ #' @noRd #' #' -sim_ar3 = function(phi, ar1, ar2, ar3, tau, last_trends, h){ +sim_ar3 = function(drift, ar1, ar2, ar3, tau, last_trends, h){ states <- rep(NA, length = h + 3) states[1] <- last_trends[1] states[2] <- last_trends[2] states[3] <- last_trends[3] for (t in 4:(h + 3)) { - states[t] <- rnorm(1, phi + ar1*states[t - 1] + + states[t] <- rnorm(1, drift + ar1*states[t - 1] + ar2*states[t - 2] + ar3*states[t - 3], sqrt(1 / tau)) } diff --git a/R/mvgam_setup.R b/R/mvgam_setup.R index 4e47d8d6..6aa01349 100644 --- a/R/mvgam_setup.R +++ b/R/mvgam_setup.R @@ -11,12 +11,11 @@ mvgam_setup <- function(formula, maxit = 40) { # Check family will work in mgcv - if (is.character(family)) - family <- eval(parse(text = family)) - if (is.function(family)) - family <- family() - if (is.null(family$family)) - stop("family not recognized") + family <- fullname_to_family(family) + + family <- match.arg(arg = family, choices = c("nb", "poisson", + "Tweedie(p = 1.5)", "betar", + "gaussian","Gamma")) # Initialise the GAM for a few iterations to get all necessary structures for # generating predictions; also provides information to regularize parametric diff --git a/R/pfilter_mvgam_fc.R b/R/pfilter_mvgam_fc.R index 0d5fcd52..d076c8fb 100644 --- a/R/pfilter_mvgam_fc.R +++ b/R/pfilter_mvgam_fc.R @@ -175,7 +175,7 @@ pfilter_mvgam_fc = function(file_path = 'pfilter', })) } else { lv_preds <- do.call(rbind, lapply(seq_len(particles[[x]]$n_lv), function(lv){ - sim_ar3(phi = particles[[x]]$phi[lv], + sim_ar3(drift = particles[[x]]$drift[lv], ar1 = particles[[x]]$ar1[lv], ar2 = particles[[x]]$ar2[lv], ar3 = particles[[x]]$ar3[lv], @@ -190,13 +190,18 @@ pfilter_mvgam_fc = function(file_path = 'pfilter', trend_preds <- as.numeric(t(lv_preds) %*% particles[[x]]$lv_coefs[series,]) Xpmat <- cbind(Xp[which(as.numeric(series_test$series) == series),], trend_preds) attr(Xpmat, 'model.offset') <- attr(Xp, 'model.offset') + + # Family-specific parameters + par_extracts <- lapply(seq_along(particles[[x]]$family_pars), function(j){ + particles[[x]]$family_pars[[j]][series] + }) + names(par_extracts) <- names(particles[[x]]$family_pars) + mvgam:::mvgam_predict(family = particles[[x]]$family, Xp = Xpmat, type = 'response', betas = c(particles[[x]]$betas, 1), - size = particles[[x]]$size[series], - p = particles[[x]]$p[series], - twdis = particles[[x]]$twdis[series]) + family_pars = par_extracts) }) @@ -209,7 +214,7 @@ pfilter_mvgam_fc = function(file_path = 'pfilter', last_trends = particles[[x]]$trend_states[[series]], h = fc_horizon) } else { - trend_preds <- sim_ar3(phi = particles[[x]]$phi[series], + trend_preds <- sim_ar3(drift = particles[[x]]$drift[series], ar1 = particles[[x]]$ar1[series], ar2 = particles[[x]]$ar2[series], ar3 = particles[[x]]$ar3[series], @@ -221,13 +226,18 @@ pfilter_mvgam_fc = function(file_path = 'pfilter', # Generate predictions on the response scale Xpmat <- cbind(Xp[which(as.numeric(series_test$series) == series),], trend_preds) attr(Xpmat, 'model.offset') <- attr(Xp, 'model.offset') + + # Family-specific parameters + par_extracts <- lapply(seq_along(particles[[x]]$family_pars), function(j){ + particles[[x]]$family_pars[[j]][series] + }) + names(par_extracts) <- names(particles[[x]]$family_pars) + mvgam:::mvgam_predict(family = particles[[x]]$family, Xp = Xpmat, type = 'response', betas = c(particles[[x]]$betas, 1), - size = particles[[x]]$size[series], - p = particles[[x]]$p[series], - twdis = particles[[x]]$twdis[series]) + family_pars = par_extracts) }) } series_fcs diff --git a/R/pfilter_mvgam_init.R b/R/pfilter_mvgam_init.R index ccae7063..fc96ae3a 100644 --- a/R/pfilter_mvgam_init.R +++ b/R/pfilter_mvgam_init.R @@ -242,14 +242,14 @@ if(object$use_lv){ # Beta coefficients for GAM component betas <- MCMCvis::MCMCchains(object$model_output, 'b') -# Phi estimates for latent trend drift terms +# drift estimates for latent trend drift terms if(object$drift){ - phis <- MCMCvis::MCMCchains(object$model_output, 'phi') + drifts <- MCMCvis::MCMCchains(object$model_output, 'drift') } else { if(object$use_lv){ - phis <- matrix(0, nrow = NROW(betas), ncol = object$n_lv) + drifts <- matrix(0, nrow = NROW(betas), ncol = object$n_lv) } else { - phis <- matrix(0, nrow = NROW(betas), ncol = n_series) + drifts <- matrix(0, nrow = NROW(betas), ncol = n_series) } } @@ -314,26 +314,15 @@ if(object$trend_model == 'AR3'){ # Family-specific parameters -if(object$family == 'Negative Binomial'){ - sizes <- MCMCvis::MCMCchains(object$model_output, 'r') -} else { - sizes <- NULL -} - -if(object$family == 'Tweedie'){ - twdis <- MCMCvis::MCMCchains(object$model_output, 'twdis') - p <- matrix(1.5, nrow = NROW(betas), ncol = n_series) -} else { - twdis <- p <- NULL -} - family <- object$family +pars <- mvgam:::extract_family_pars(family = family, + object = object) # Generate sample sequence for n_particles -if(n_particles < dim(phis)[1]){ - sample_seq <- sample(seq_len(dim(phis)[1]), size = n_particles, replace = F) +if(n_particles < dim(drifts)[1]){ + sample_seq <- sample(seq_len(dim(drifts)[1]), size = n_particles, replace = F) } else { - sample_seq <- sample(seq_len(dim(phis)[1]), size = n_particles, replace = T) + sample_seq <- sample(seq_len(dim(drifts)[1]), size = n_particles, replace = T) } #### 2. Generate particles and calculate their proposal weights #### @@ -353,7 +342,7 @@ clusterExport(NULL, c('use_lv', 'truth', 'last_assim', 'taus', - 'phis', + 'drifts', 'ar1s', 'ar2s', 'ar3s', @@ -363,9 +352,7 @@ clusterExport(NULL, c('use_lv', 'lv_coefs', 'trends', 'family', - 'sizes', - 'twdis', - 'p', + 'pars', 'n_series', 'n_particles', 'upper_bounds'), @@ -379,7 +366,7 @@ particles <- pbapply::pblapply(sample_seq, function(x){ if(is.null(alpha_gps)){ # Sample AR parameters - phi <- phis[samp_index, ] + drift <- drifts[samp_index, ] ar1 <- ar1s[samp_index, ] ar2 <- ar2s[samp_index, ] ar3 <- ar3s[samp_index, ] @@ -388,7 +375,7 @@ particles <- pbapply::pblapply(sample_seq, function(x){ # Sample trend precisions tau <- taus[samp_index,] } else { - phi <- ar1 <- ar2 <- ar3 <- tau <- 0 + drift <- ar1 <- ar2 <- ar3 <- tau <- 0 alpha_gp <- alpha_gps[samp_index, ] rho_gp <- rho_gps[samp_index, ] } @@ -396,11 +383,6 @@ particles <- pbapply::pblapply(sample_seq, function(x){ # Sample beta coefs betas <- betas[samp_index, ] - # Family-specific parameters - size <- sizes[samp_index, ] - twdis <- twdis[samp_index, ] - p <- p[samp_index] - if(use_lv){ # Sample a last state estimate for the latent variables @@ -417,7 +399,7 @@ particles <- pbapply::pblapply(sample_seq, function(x){ if(is.null(alpha_gps)){ lv_states <- do.call(cbind, lapply(seq_along(lvs), function(lv){ sim_ar3(h = 1, - phi = phi[lv], + drift = drift[lv], ar1 = ar1[lv], ar2 = ar2[lv], ar3 = ar3[lv], @@ -457,26 +439,29 @@ particles <- pbapply::pblapply(sample_seq, function(x){ if(is.na(truth[series])){ weight <- 1 } else { - if(family == 'Negative Binomial'){ - weight <- (dnbinom(truth[series], size = size[series], - mu = exp(((Xp[which(as.numeric(series_test$series) == series),] %*% betas)) + - (trend_states[series])))) - } - if(family == 'Poisson'){ - weight <- (dpois(truth[series], - lambda = exp(((Xp[which(as.numeric(series_test$series) == series),] %*% betas)) + - (trend_states[series])))) - } + # Family-specific parameters + par_extracts <- lapply(seq_along(pars), function(x){ + if(is.matrix(pars[[x]])){ + pars[[x]][samp_index, series] + } else { + pars[[x]][samp_index] + } + }) + names(par_extracts) <- names(pars) - if(family == 'Tweedie'){ - weight <- exp(mgcv::ldTweedie(y = truth[series], - mu = exp(((Xp[which(as.numeric(series_test$series) == series),] %*% betas)) + - (trend_states[series])), - p = p, - phi = twdis[series], - all.derivs = F)[,1]) - } + # Series-specific linear predictor matrix + Xpmat <- t(rbind(as.matrix(Xp[which(as.numeric(series_test$series) == series),]), + trend_states[series])) + attr(Xpmat, 'model.offset') <- attr(Xp, 'model.offset') + + # Likelihood + weight <- mvgam:::mvgam_predict(family = family, + family_pars = par_extracts, + truth = truth[series], + Xp = Xpmat, + betas = c(betas, 1), + density = TRUE) } weight @@ -499,7 +484,7 @@ particles <- pbapply::pblapply(sample_seq, function(x){ if(is.null(alpha_gps)){ trends <- do.call(cbind, lapply(seq_along(trends), function(trend){ sim_ar3(h = 1, - phi = phi[trend], + drift = drift[trend], ar1 = ar1[trend], ar2 = ar2[trend], ar3 = ar3[trend], @@ -521,26 +506,29 @@ particles <- pbapply::pblapply(sample_seq, function(x){ if(is.na(truth[series])){ weight <- 1 } else { - if(family == 'Negative Binomial'){ - weight <- (dnbinom(truth[series], size = size[series], - mu = exp(((Xp[which(as.numeric(series_test$series) == series),] %*% betas)) + - (trends[series])))) - } - if(family == 'Poisson'){ - weight <- (dpois(truth[series], - lambda = exp(((Xp[which(as.numeric(series_test$series) == series),] %*% betas)) + - (trends[series])))) - } - - if(family == 'Tweedie'){ - weight <- exp(mgcv::ldTweedie(y = truth[series], - mu = exp(((Xp[which(as.numeric(series_test$series) == series),] %*% betas)) + - (trends[series])), - p = p, - phi = twdis[series], - all.derivs = F)[,1]) - } + # Family-specific parameters + par_extracts <- lapply(seq_along(pars), function(x){ + if(is.matrix(pars[[x]])){ + pars[[x]][samp_index, series] + } else { + pars[[x]][samp_index] + } + }) + names(par_extracts) <- names(pars) + + # Series-specific linear predictor matrix + Xpmat <- t(rbind(as.matrix(Xp[which(as.numeric(series_test$series) == series),]), + trends[series])) + attr(Xpmat, 'model.offset') <- attr(Xp, 'model.offset') + + # Likelihood + weight <- mvgam:::mvgam_predict(family = family, + family_pars = par_extracts, + truth = truth[series], + Xp = Xpmat, + betas = c(betas, 1), + density = TRUE) } weight @@ -562,6 +550,16 @@ particles <- pbapply::pblapply(sample_seq, function(x){ names(trend_states) <- paste0('series', seq_len(n_series)) } + # Family-specific parameters + family_pars <- lapply(seq_along(pars), function(x){ + if(is.matrix(pars[[x]])){ + pars[[x]][samp_index, ] + } else { + pars[[x]][samp_index] + } + }) + names(family_pars) <- names(pars) + # Store important particle-specific information for later filtering list(use_lv = use_lv, n_lv = n_lv, @@ -570,16 +568,14 @@ particles <- pbapply::pblapply(sample_seq, function(x){ lv_states = lv_states, lv_coefs = lv_coefs, betas = as.numeric(betas), - phi = as.numeric(phi), + drift = as.numeric(drift), ar1 = as.numeric(ar1), ar2 = as.numeric(ar2), ar3 = as.numeric(ar3), tau = as.numeric(tau), alpha_gp = as.numeric(alpha_gp), rho_gp = as.numeric(rho_gp), - size = as.numeric(size), - p = as.numeric(p), - twdis = as.numeric(twdis), + family_pars = family_pars, trend_states = lapply(trend_states, unname), weight = weight, liks = liks, diff --git a/R/pfilter_mvgam_smooth.R b/R/pfilter_mvgam_smooth.R index e3eb3a32..bb02e6ae 100644 --- a/R/pfilter_mvgam_smooth.R +++ b/R/pfilter_mvgam_smooth.R @@ -104,7 +104,7 @@ pfilter_mvgam_smooth = function(particles, lv_states <- unlist(lapply(seq_len(length(particles[[x]]$lv_states)), function(lv){ sim_ar3(h = 1, last_trends = particles[[x]]$lv_states[[lv]][1:3], - phi = 0, + drift = 0, ar1 = particles[[x]]$ar1[lv], ar2 = particles[[x]]$ar2[lv], ar3 = particles[[x]]$ar3[lv], @@ -148,7 +148,7 @@ pfilter_mvgam_smooth = function(particles, next_trend <- unlist(lapply(seq_len(length(particles[[x]]$trend_states)), function(series){ sim_ar3(h = 1, last_trends = particles[[x]]$trend_states[[series]][1:3], - phi = particles[[x]]$phi[series], + drift = particles[[x]]$drift[series], ar1 = particles[[x]]$ar1[series], ar2 = particles[[x]]$ar2[series], ar3 = particles[[x]]$ar3[series], @@ -180,32 +180,47 @@ pfilter_mvgam_smooth = function(particles, if(is.na(next_assim$y[series])){ series_weight <- 1 } else { - if(particles[[x]]$family == 'Negative Binomial'){ - series_weight <- (dnbinom(next_assim$y[series], - size = particles[[x]]$size[series], - mu = exp(((Xp[which(as.numeric(next_assim$series) == series),] %*% - particles[[x]]$betas)) + - (next_trend[series]) + - attr(Xp, 'model.offset')))) - } - if(particles[[x]]$family == 'Poisson'){ - series_weight <- (dpois(next_assim$y[series], - lambda = exp(((Xp[which(as.numeric(next_assim$series) == series),] %*% - particles[[x]]$betas)) + - (next_trend[series]) + - attr(Xp, 'model.offset')))) - } - - if(particles[[x]]$family == 'Tweedie'){ - series_weight <- exp(mgcv::ldTweedie(y = next_assim$y[series], - mu = exp(((Xp[which(as.numeric(next_assim$series) == series),] %*% - particles[[x]]$betas)) + (next_trend[series]) + - attr(Xp, 'model.offset')), - p = particles[[x]]$p, - phi = particles[[x]]$twdis[series], - all.derivs = F)[,1]) - } + # Series-specific linear predictor matrix + Xpmat <- t(rbind(as.matrix(Xp[which(as.numeric(next_assim$series) == series),]), + next_trend[series])) + attr(Xpmat, 'model.offset') <- attr(Xp, 'model.offset') + + # Likelihood + series_weight <- mvgam:::mvgam_predict(family = particles[[x]]$family, + family_pars = particles[[x]]$family_pars, + truth = next_assim$y[series], + Xp = Xpmat, + betas = c(particles[[x]]$betas, 1), + density = TRUE) + + + # if(particles[[x]]$family == 'Negative Binomial'){ + # series_weight <- (dnbinom(next_assim$y[series], + # size = particles[[x]]$size[series], + # mu = exp(((Xp[which(as.numeric(next_assim$series) == series),] %*% + # particles[[x]]$betas)) + + # (next_trend[series]) + + # attr(Xp, 'model.offset')))) + # } + # + # if(particles[[x]]$family == 'Poisson'){ + # series_weight <- (dpois(next_assim$y[series], + # lambda = exp(((Xp[which(as.numeric(next_assim$series) == series),] %*% + # particles[[x]]$betas)) + + # (next_trend[series]) + + # attr(Xp, 'model.offset')))) + # } + # + # if(particles[[x]]$family == 'Tweedie'){ + # series_weight <- exp(mgcv::ldTweedie(y = next_assim$y[series], + # mu = exp(((Xp[which(as.numeric(next_assim$series) == series),] %*% + # particles[[x]]$betas)) + (next_trend[series]) + + # attr(Xp, 'model.offset')), + # p = particles[[x]]$p, + # drift = particles[[x]]$twdis[series], + # all.derivs = F)[,1]) + # } } series_weight @@ -222,14 +237,12 @@ pfilter_mvgam_smooth = function(particles, lv_states = lv_states, lv_coefs = particles[[x]]$lv_coefs, betas = particles[[x]]$betas, - phi = as.numeric(particles[[x]]$phi), + drift = as.numeric(particles[[x]]$drift), ar1 = as.numeric(particles[[x]]$ar1), ar2 = as.numeric(particles[[x]]$ar2), ar3 = as.numeric(particles[[x]]$ar3), tau = as.numeric(particles[[x]]$tau), - size = particles[[x]]$size, - p = particles[[x]]$p, - twdis = particles[[x]]$twdis, + family_pars = particles[[x]]$family_pars, alpha_gp = particles[[x]]$alpha_gp, rho_gp = particles[[x]]$rho_gp, trend_states = trend_states, @@ -241,28 +254,6 @@ pfilter_mvgam_smooth = function(particles, }, cl = cl) stopCluster(cl) - # n_series <- length(particles[[1]]$trend_states) - # if(n_series > 1){ - # lik_weights <- do.call(rbind, lapply(seq_along(particles), function(x){ - # particles[[x]]$liks - # })) - # - # series_sds <- apply(lik_weights, 2, function(x) sd(x)) - # if(all(series_sds == 0)){ - # # Do nothing if all series have equal weights due to resampling or all NAs in - # # last observation - # } else { - # # Else update weights using ranking information - # lik_weights <- apply(apply(lik_weights[, !series_sds == 0 ], 2, rank), 1, prod) - # lik_weights <- lik_weights / max(lik_weights) - # - # particles <- lapply(seq_along(particles), function(x){ - # particles[[x]]$weight <- lik_weights[x] * particles[[x]]$weight - # particles[[x]] - # }) - # } - # } - weights <- (unlist(lapply(seq_along(particles), function(x){ tail(particles[[x]]$weight, 1)}))) @@ -624,20 +615,18 @@ pfilter_mvgam_smooth = function(particles, lv_states = lv_evolve, lv_coefs = lv_coefs_evolve, betas = betas, - size = particles[[x]]$size, tau = if(use_resampling){ particles[[x]]$tau } else { pmin(100, particles[[x]]$tau + runif(length(particles[[x]]$tau), 0, evolve)) }, - phi = particles[[x]]$phi, + drift = particles[[x]]$drift, ar1 = particles[[x]]$ar1, ar2 = particles[[x]]$ar2, ar3 = particles[[x]]$ar3, - p = particles[[x]]$p, - twdis = particles[[x]]$twdis, alpha_gp = particles[[x]]$alpha_gp, rho_gp = particles[[x]]$rho_gp, + family_pars = particles[[x]]$family_pars, trend_states = trend_evolve, weight = particle_weight, upper_bounds = particles[[x]]$upper_bounds, diff --git a/R/predict.mvgam.R b/R/predict.mvgam.R index 04cbd70b..974f9522 100644 --- a/R/predict.mvgam.R +++ b/R/predict.mvgam.R @@ -68,21 +68,9 @@ predict.mvgam = function(object, newdata, data_test, type = 'link', # Family of model family <- object$family - # Negative binomial size estimate - if(family == 'Negative Binomial'){ - sizes <- MCMCvis::MCMCchains(object$model_output, 'r') - } else { - sizes <- NULL - } - - # Tweedie parameters - if(family == 'Tweedie'){ - twdiss <- MCMCvis::MCMCchains(object$model_output, 'twdis') - ps <- matrix(1.5, nrow = NROW(betas), ncol = NCOL(object$ytimes)) - } else { - twdiss <- NULL - ps <- NULL - } + # Family-specific parameters + pars <- mvgam:::extract_family_pars(family = family, + object = object) # Determine which series each observation belongs to series_ind <- as.numeric(newdata$series) @@ -91,9 +79,7 @@ predict.mvgam = function(object, newdata, data_test, type = 'link', cl <- parallel::makePSOCKcluster(n_cores) setDefaultCluster(cl) clusterExport(NULL, c('betas', - 'sizes', - 'twdiss', - 'ps', + 'pars', 'newdata', 'Xp', 'series_ind'), @@ -101,13 +87,20 @@ predict.mvgam = function(object, newdata, data_test, type = 'link', pbapply::pboptions(type = "none") predictions <- do.call(rbind, pbapply::pblapply(seq_len(dim(betas)[1]), function(x){ - mvgam_predict(family = family, + # Family-specific parameters + par_extracts <- lapply(seq_along(pars), function(j){ + if(is.matrix(pars[[j]])){ + pars[[j]][x,series_ind[x]] + } else { + pars[[j]][x] + } + }) + names(par_extracts) <- names(pars) + mvgam:::mvgam_predict(family = family, Xp = Xp, type = type, betas = betas[x,], - size = sizes[x, series_ind], - p = ps[x, series_ind], - twdis = twdiss[x, series_ind]) + family_pars = par_extracts) }, cl = cl)) stopCluster(cl) diff --git a/R/print.mvgam.R b/R/print.mvgam.R index 1f50c8fa..675b2665 100644 --- a/R/print.mvgam.R +++ b/R/print.mvgam.R @@ -25,7 +25,7 @@ cat(paste0(object$family, '\n')) message() message("Link function:") -cat(paste0('log', '\n')) +cat(paste0(family_links(object$family), '\n')) message() message("Trend model:") @@ -58,25 +58,25 @@ message() if(object$fit_engine == 'jags'){ message('Status:') - cat('Fitted using runjags::run.jags()', '\n') + cat('Fitted using JAGS', '\n') message() } if(object$fit_engine == 'stan'){ message('Status:') - cat('Fitted using rstan::stan()', '\n') + cat('Fitted using Stan', '\n') message() } if(object$family == 'Negative Binomial'){ message("Dispersion parameter estimates:") - print(MCMCvis::MCMCsummary(object$model_output, 'r')[,c(3:7)]) + print(MCMCvis::MCMCsummary(object$model_output, 'phi')[,c(3:7)]) message() } if(object$family == 'Tweedie'){ message("Dispersion parameter estimates:") - print(MCMCvis::MCMCsummary(object$model_output, 'twdis')[,c(3:7)]) + print(MCMCvis::MCMCsummary(object$model_output, 'phi')[,c(3:7)]) message() } @@ -99,7 +99,7 @@ print.mvgam_prefit = function(object){ message() message("Link function:") - cat(paste0('log', '\n')) + cat(paste0(family_links(object$family), '\n')) message() message("Trend model:") diff --git a/R/summary.mvgam.R b/R/summary.mvgam.R index 6d264076..f05adf0d 100644 --- a/R/summary.mvgam.R +++ b/R/summary.mvgam.R @@ -12,7 +12,7 @@ #'@export summary.mvgam = function(object){ -#### Standard summary of formula and model argumements #### +#### Standard summary of formula and model arguments #### message("GAM formula:") print(object$call) message() @@ -22,7 +22,7 @@ cat(paste0(object$family, '\n')) message() message("Link function:") -cat(paste0('log', '\n')) +cat(paste0(family_links(object$family), '\n')) message() message("Trend model:") @@ -67,13 +67,13 @@ if(object$fit_engine == 'stan'){ if(object$family == 'Negative Binomial'){ message("Dispersion parameter estimates:") - print(MCMCvis::MCMCsummary(object$model_output, 'r')[,c(3:7)]) + print(MCMCvis::MCMCsummary(object$model_output, 'phi')[,c(3:7)]) message() } if(object$family == 'Tweedie'){ message("Dispersion parameter estimates:") - print(MCMCvis::MCMCsummary(object$model_output, 'twdis')[,c(3:7)]) + print(MCMCvis::MCMCsummary(object$model_output, 'phi')[,c(3:7)]) message() } @@ -156,8 +156,8 @@ if(object$use_lv){ if(object$trend_model != 'None'){ if(object$trend_model == 'RW'){ if(object$drift){ - message("Latent trend drift (phi) estimates:") - print(MCMCvis::MCMCsummary(object$model_output, c('phi'))[,c(3:7)]) + message("Latent trend drift estimates:") + print(MCMCvis::MCMCsummary(object$model_output, c('drift'))[,c(3:7)]) message() } else { } @@ -171,8 +171,8 @@ if(object$use_lv){ if(object$trend_model == 'AR1'){ if(object$drift){ - message("Latent trend drift (phi) and parameter estimates:") - print(MCMCvis::MCMCsummary(object$model_output, c('phi', 'ar1'))[,c(3:7)]) + message("Latent trend drift and parameter estimates:") + print(MCMCvis::MCMCsummary(object$model_output, c('drift', 'ar1'))[,c(3:7)]) message() } else { message("Latent trend parameter estimates:") @@ -183,8 +183,8 @@ if(object$use_lv){ if(object$trend_model == 'AR2'){ if(object$drift){ - message("Latent trend drift (phi) and parameter estimates:") - print(MCMCvis::MCMCsummary(object$model_output, c('phi', 'ar1', 'ar2'))[,c(3:7)]) + message("Latent trend drift and parameter estimates:") + print(MCMCvis::MCMCsummary(object$model_output, c('drift', 'ar1', 'ar2'))[,c(3:7)]) message() } else { message("Latent trend parameter estimates:") @@ -195,8 +195,8 @@ if(object$use_lv){ if(object$trend_model == 'AR3'){ if(object$drift){ - message("Latent trend drift (phi) and parameter estimates:") - print(MCMCvis::MCMCsummary(object$model_output, c('phi', 'ar1', 'ar2', 'ar3'))[,c(3:7)]) + message("Latent trend drift and parameter estimates:") + print(MCMCvis::MCMCsummary(object$model_output, c('drift', 'ar1', 'ar2', 'ar3'))[,c(3:7)]) message() } else { message("Latent trend parameter estimates:") @@ -211,8 +211,8 @@ if(!object$use_lv){ if(object$trend_model != 'None'){ if(object$trend_model == 'RW'){ if(object$drift){ - message("Latent trend drift (phi) and sigma estimates:") - print(MCMCvis::MCMCsummary(object$model_output, c('phi', 'sigma'))[,c(3:7)]) + message("Latent trend drift and sigma estimates:") + print(MCMCvis::MCMCsummary(object$model_output, c('drift', 'sigma'))[,c(3:7)]) message() } else { message("Latent trend variance estimates:") @@ -223,8 +223,8 @@ if(!object$use_lv){ if(object$trend_model == 'AR1'){ if(object$drift){ - message("Latent trend drift (phi) and parameter estimates:") - print(MCMCvis::MCMCsummary(object$model_output, c('phi', 'ar1', 'sigma'))[,c(3:7)]) + message("Latent trend drift and parameter estimates:") + print(MCMCvis::MCMCsummary(object$model_output, c('drift', 'ar1', 'sigma'))[,c(3:7)]) message() } else { message("Latent trend parameter estimates:") @@ -235,8 +235,8 @@ if(!object$use_lv){ if(object$trend_model == 'AR2'){ if(object$drift){ - message("Latent trend drift (phi) and parameter estimates:") - print(MCMCvis::MCMCsummary(object$model_output, c('phi', 'ar1', 'ar2', 'sigma'))[,c(3:7)]) + message("Latent trend drift and parameter estimates:") + print(MCMCvis::MCMCsummary(object$model_output, c('drift', 'ar1', 'ar2', 'sigma'))[,c(3:7)]) message() } else { message("Latent trend parameter estimates:") @@ -247,8 +247,8 @@ if(!object$use_lv){ if(object$trend_model == 'AR3'){ if(object$drift){ - message("Latent trend drift (phi) and parameter estimates:") - print(MCMCvis::MCMCsummary(object$model_output, c('phi', 'ar1', 'ar2', 'ar3', 'sigma'))[,c(3:7)]) + message("Latent trend drift and parameter estimates:") + print(MCMCvis::MCMCsummary(object$model_output, c('drift', 'ar1', 'ar2', 'ar3', 'sigma'))[,c(3:7)]) message() } else { message("Latent trend parameter estimates:") diff --git a/R/vectorise_stan_lik.R b/R/vectorise_stan_lik.R index 6f84287a..10352b6b 100644 --- a/R/vectorise_stan_lik.R +++ b/R/vectorise_stan_lik.R @@ -107,25 +107,25 @@ vectorise_stan_lik = function(model_file, model_data, family = 'Poisson', if(threads > 1){ model_file[lik_line] <- paste0('{\n// likelihood functions\n', 'vector[n_nonmissing] flat_trends;\n', - 'real flat_rs[n_nonmissing];\n', + 'real flat_phis[n_nonmissing];\n', 'flat_trends = (to_vector(trend))[obs_ind];\n', - 'flat_rs = to_array_1d(rep_each(r_inv, n)[obs_ind]);\n', + 'flat_phis = to_array_1d(rep_each(phi_inv, n)[obs_ind]);\n', 'target += reduce_sum(partial_log_lik_lpmf, seq,\n', 'grainsize,\n', 'flat_ys,\n', ifelse(offset,'exp(append_col(flat_xs, flat_trends) * append_row(b, 1.0)) + offset[obs_ind],\n', 'exp(append_col(flat_xs, flat_trends) * append_row(b, 1.0)),\n'), - 'flat_rs);\n}\n}\n') + 'flat_phis);\n}\n}\n') } else { model_file[lik_line] <- paste0('{\n// likelihood functions\n', 'vector[n_nonmissing] flat_trends;\n', - 'real flat_rs[n_nonmissing];\n', + 'real flat_phis[n_nonmissing];\n', 'flat_trends = (to_vector(trend))[obs_ind];\n', - 'flat_rs = to_array_1d(rep_each(r_inv, n)[obs_ind]);\n', + 'flat_phis = to_array_1d(rep_each(phi_inv, n)[obs_ind]);\n', 'flat_ys ~ neg_binomial_2(\n', ifelse(offset,'exp(append_col(flat_xs, flat_trends) * append_row(b, 1.0)) + offset[obs_ind],\n', 'exp(append_col(flat_xs, flat_trends) * append_row(b, 1.0)),\n'), - 'inv(flat_rs));\n}\n}\n') + 'inv(flat_phis));\n}\n}\n') } if(any(grepl('functions {', model_file, fixed = TRUE))){ @@ -260,7 +260,7 @@ vectorise_stan_lik = function(model_file, model_data, family = 'Poisson', 'eta = X * b;\n'), 'for(s in 1:n_series){ \n', 'mus[1:n, s] = eta[ytimes[1:n, s]] + trend[1:n, s];\n', - 'ypred[1:n, s] = neg_binomial_2_rng(exp(mus[1:n, s]), r_vec[1:n, s]);\n', + 'ypred[1:n, s] = neg_binomial_2_rng(exp(mus[1:n, s]), phi_vec[1:n, s]);\n', '}') } diff --git a/docs/README-unnamed-chunk-11-1.png b/docs/README-unnamed-chunk-11-1.png index 501d5431eda893c6b44311a393572d64de6ad129..27db2d3077c61110ffd0967c729b2cef3ed87496 100644 GIT binary patch literal 17071 zcmdtK1ytKl*DgwfyA*d?+})j0+}(qFakr*{Qlw~cD8*acgC}@#cb7tOiaR&`zxTc8 zTj#F3&icM}?pf=c$^0hC?Ad!}&rE(ZnLJOTG}RTbF~~6x5D>7H6yIngARxlv7Xlh0 zJZE|Ri346DJ5f7$Ob~_VC&1&9 z9lZ2E;(r7Qe}#WT0{mf@08bn3;AI*c?Hb|fb1(uzf+#$SCfLE_Kk@{4XoNo!8sX_u z!ct?x58jsNNM1|aU(#|wNDuw%ebD{5shPX34eP0`R`Au* za)kaGxrOtIxq;{H{V(}Xut>%aFj@^z+3)-^ei2kA^-$f_vhvDtNkygiusLKTlxB=z zo*Hgq*7#$#6TLNt-p9i?0T2-E+pz9>5za}eLs_Ut`0zU>`)7-hdISKCgb{`*)hY)! zSdcio0Cqw{ghx8PO?!IZpF+=666PFfP4xa=`g^D$)W?5!)FABA?cw%lcg8*R^cRIm zD(Y1gwQiP=Qme~0& z7+}Eb9KsB~|4tKM(JSS#T8Rl=ZoiBX|nQFhBiF{X<{46eH!|>5W9w?A} zcOF(1m&9iH6^Fiu(ISH69~beBu}AJ3aS!B+QW+;A5{esP)hT!l ziwq8 zOeD+uj|Om$#Aec~P{WAu)_lOL)p}l&ztmdqrqVpoY`RQ}_O?s$C@)pXu{FLpi!(Q* zLfR5nQwhjSi=2udJ%|nN7Z{StrmXGXhKP%a!0gbci|e(pVw_KH}2r zJ56WDYaiteWC)MsECVapg+uOcJtlA*B~(1BA$b*k)d324hLVo-WA9Q_-&|9*+LSib z+ZueL%j@9#7y?ij+63Un@RvwAOc@@@d3?4{2S44e?5zV&aiQ0ew&HCeW!R}vY?;$8E{`I>r9OAu6sezqFpM?9!;%o4qYiPL8>b2j`eG>sD>kGCTeo&5q)m}zA z$E)c@#_F%*)qHX+csIX|OxI03u$y_5n=)mdPwif}>3rUG=23c#vn)F+IvuUnu0Rjg z`riL-Q>!L5YT=b~1qa{&T}@2KI0mJ^ZxfJq?{>cta0Wf_E0tNsQG4t@+v3hYnT4GC z80S0$`6Y&1vTdy>X2KR%6bEZl|GcB`cBGg6zin7^kyDZlVVdSJI=BiF>VWn-E8Z%NM(WoH>S9!BW+RNYVe34ShfM_91aV~Z z-_7H5Xt4KW0T`wc>$fBh!_Z*_=|1>f@D=p4cb$>VCXTdl1_l(UGp_aQyd9}_CpUqNTI7+te{s8i1130_4FzD6AX{^{hm??)`QRW zs1JyRA3-Nsbv<7=;OfrK>#Qwc%Z{(!}= zdGcU7?cO}5(EbCrhNgXA_pIQv9dy5f8;-b60QUm*_$}Z)Pu;!F_hGXa2>M8`xDjGO zFa&NC03wVO4GY1Vz&xtxgLB@C^h{j?s~mXic9=4x(%|bbNtA^LA0oobML9z<*+gqCQ$_x##r000F>c z%T(v>1KS6|SQZ@F&0be~(z^57I9hV2Bcf%aC0A*Zs$wIlgBe!4fZQB-!pK*enBdMr z>2)Mwj)*RxEUi0WIT|ja!V~_=BLmmMLZif#*C54W1OOCdF{`*qRS}Wmz`_!J?Bsu6 zeK=;oLeW70S)bMr0Kn+)VLUgAQFvI42mpokU3x5(JrHc;h8!I%18)ZRlVle#fgFue z`O{}SQhF2sDgmJy{s$Q=T6aL3kXK{rbjCrm_UF= zDg7lJEFkLqJf71h{#WWPc&#Gt?Bda2VHR~(90eTxcz9|r)8JuRw*p5H7Qp0 z&_cp7^>zBQPGG_(-6vDR>%y(o#{iF9ZE*$A)HVPt10cBdy4S>R63QA2eN8x~wywJ) z1560F0hG1upOD7bGrN}t1Gzt;;?Ku;ufnZ-@a~6Q$FGwL7x}YP`?i;V(iCb+GWvVg#*0#HCQSOX8u?9o9`Ma{|`ApNPgNH-8qRWu9- zCHe2%9FtpT;ehwwhwvxU1jZ5NKQ2gLdEH2{WR-s7{vif7`x65+4oI<3HVQ9xGktNZ zb;VIZC4yK0FwmXaKl=x=92!ce^tx8@H6{S8NqfpXbr6AvB@2QT3I9lQWL* zVb;U1NRUQz_+F|Ql!n8$hBt|2-B(F3M`W$g#Jg0TWY&XuS~H_M@#aTN^YH0}O%A{A ztaBsz@?{y|3G2KiG&8nDJGAM5eZtFT*z)D1jG(JdCad_+NhYh~rS$iWNYMvav&^-p z2$-%q#;D+xtoyqg#pW+vc=MgZ$};P1r~IB7QT#~0+;v)Logy1$fMRu+$Y8!# zJ-x_SA7Q>bbLaqjx$jY4Y*?TQc064C4_bo=zv|B*zj{|3^^c0OcN5Q zj#e8c0v8ddW{gJQK86A5zMx2*Hm+2Eu_Hog#JXg zUFI=;p)FdJA7Lt;qib+QdMursi|CF;yFb6c)79r`AQ&dIss$$w-)Ad#=|)=O&==_u zFfXb1X~nkH9bL1HBk&49xlmOk&Ry^Hz)B`($ch9mv!ok^jug`KX#A|K|H;xLzNh`0aMk}?`>4T-H<8}+`M&kynCuxI{mZ= zJ}s5wRvI(nu-oUN0jX(StBNr)x=4I=XQ#0GJYeW)R@T{}K3c%76DERon8}?xK~=n> z%6)v|5b{`~>--r7bx=5fc8C1^Iq`f^<@+#NUq*MgLrz-4vd$?&2wozA`VVe{Ff#PR zQRc*M>!+6UxK7z`qky!hAa$?##oF1~ujUqBbZk7rHJ$@R>Ui&cTOU^X zx%)M>GB|7%YdgRhKa9jnI@r>>(%2fK_Iw?x*SIfjq7)=vBdqNCCPZg?`7u%d!Fw#C z_%{RO@w9|~w+X5wAV`+YxVw#4JKJEb=M@HWI^T@Vt$a_FTtix8H_t--} z=IZCMx_FM4bQ2i20l>Sp2Fx7gO*I{Z`YR}ru*n$w%=-VfS9qX=e&*l*5**9cakAw6 zk|ld@y@uD~;xlKW_ooz(d)%kCz`KjROG6^Z}f7bmsn!FL6517 zcwU13z9A9X5-~oPT2EK0`7j}hk&3E;1^$-fu}C!4n6>jq(YzU^EEf+6l9OuF(A5G( ztEW>6^Xm36;h!dSH|J-a;IWv^QE=Rp8=BX8>Jw>`-&4y^JB$^pNu}|_ah-h+d)+k# zZDFI39V@z^$l4GLh&K*NkjgGkH7*15mL*9)#UDHsR-r0~@)StsVs!J}c!Az6j$r8_&!U?SzE2|Ra)aK3hzIORZr)c` z+4SXw?WZY;u`gP-f4xnN=n&}1bdZ?bD?`t!>dus+fj-h)qUUX@Xrq1R2t@~pWvn(% zpWv8TA9mlUWBU1{bs2G$Z*RFNi4*Dv%0Ud)=*1%S!Z@>(+}oKigFkMjWys=QIayTt zhFRZpSm_0x3YvgPEZVhcRSSzM$9G1rj%qy(|LoqayM|o-p1N1)xyRp(wy?o3n}q~c zEICGd?9rQ7b!P0D5McSC0VCnjPPD zdv}njh-rX;bZ4@I_pyQH!hKn#m6W&kCJH%}?X88<>i(4etE1DTb@%dK9<8QfRltLcKJCDt_E)$ zesJHpOHHL+ZVXdZWp_|q~84H!@QrxlOmy_lj6IQPR zxnLVEAGj&PR2nMVp_gU+n;h2M_o%IIBb1VO1yv#Ut4Le+)3lU%?!$Sf`6ynKkS}Fr zk!!9=yu|M5WiVp5rP;Elqe~7*T=2emz2o==&&c|x)h&^DyB#-{;exRqJC*X~WgLp) z@wHjp?v}^gfYsr3Zb+KJ+1x@t7-|=R0%pM&+jlQJl||t*YERmmE$}OavS6a;#vtc$ zShc&H3_?SMyNkxD0j=43FR`bz6&5pB5wfJ!G?KA7U6DWIdoZ_!cTrw)Fo2i*7`0s0 z|4!ZL&Ew}%YP5E#5{XQS+8YlL`YP+dw7=b&Ch5};ES%3>8*_FeZrnqsOOxcw9ht+m z3zW0RZ$%B4B&n%YvY$n z6Mo)oeNv;6W$Oy8`PLmaUb@#T2u-qh7fMznKuqRuZ|VYUd&?gn1k zeRNt}#%OyO!OG0#+~??H%AZ6TTuugZ>uqS46Xi6?S&nsms5VU{^|k!HT(34mg%LUU zi-)=7YR+FsTM&?(RjQ&+@#ax~v8Ev4Tv9!VjB>|)JFj}qmk#aet4MkqwT$;c%E-GL zZyYnDt)^@|1*EH6kGl0u^s@0$VuGAAQt_OQ{Z9}ZS~67w;qMe+!r!q~VYPbe0tN35`Oy zYQe-@k<%#7bqQuxnh(S<{?|F*%!en{^4-s{8T5_V+^C|}<6SW;-*%`|WRv|w2TG*b zUUEriDfH@#i0JYK(4e4LM|H?!fsM^t7&mx#U}yVTT;#hz2m%hVu=ZLl#bGfuc~s{r z#u!C76QAL%pLmEJ&h--x-~fWbM*xvK-?a@B+c0QeRsX)?LPHu#@w%b1b}Ya5tlf!{ zoAd+9b9kCJ+EWr_pnpAQ!eRmlUmt%z1qWW<%9M%KiO)$W0$mvf{1O~W%%;`rM)P~g z{bxjgy4yCVMq;Cndc$%Xl}5T@L)xN6Jt`|7S92fBpksI*4Vxjzvot{5d&wKn=xXF_cu^@{-Fjy8{VwF z$4nb0!8KTE(b0mWux{zNlSw9Cu`%Ry+Jieh3WBsE1Pt+-#6bMcf4o81 zV|{2jR>stwc>`ri<$dd_VD}`sjZkFPk$LlB;K-UNLRn;+foZ|%`j~uPVtGYXC)o+j!=u@r>@K+3g1PZ;{?Qq%J%{ed}hQx?~e|Vt2kZzu`Vx z6_c$(gK)FwqK+Q&Cj}g5cUf%oCv0CIrH7!2;Sp0&dl=~ZPfl zHUgz}8s(z;-W(Xk6`D(t5Xb7eR**D?T_mbhIK{kGwLL2Xyr!6EW-7>w?zROJj|N6J8 z=6t=(aMK)t8fo}P&)^ZN6`E_x!Ci+}QQfIzHrlmt{rPkMu%R5n;QqU7Ri7pmQJ0Mc zg)Y1caV~Z;b^*DbZ$k58kJeS69zPpqL6g4wpH$9QpcbKmKKi6QGOUr#$nFldifaQ0oCyet+vLc#O^{Q zRZDkK$eZ-mxfHG%Vp+NDfiMb!>&tzo?;`9axhI6m%j2^vL=-an%zxB;tgRdk3>ZI~ zT}9&>W34J-)JkRKyd~Fs%i4S%VUE#Yz^Ow@{y0BTY)5_E7*2_C-P{tBh+UaG=%A6K zyOx7(sbO_~#5YN9$rh{&O(mb7+IT9}))79=^eDDvBg+*iWQi_8V@JlKiRz{1T`Id) zp5uFqo>E}Baj@pjIe3uOhiKgq=eu|CdV21+du?5=z84iMs%?4bJEsrk4wsmdMwni^ zS&=GhR&`4zwKoA;uHUp6s6|UoejJCXXiVo=YR#=zJ5+b%%eo=Q|B;xMUGq>0DT^&6 zPu=J7t_vUt**gvtBz6T?2vXrctT;)q*x6I+E)vVhiM;vxawhj7)Lhp|^E@&x`L4+F z`0(bH0EV+73mLm)C!+8I85O8r=eVogcU z$?>?LkV@7{zTX68;F*s8(4R95cBJDtzJ)&adYiLroE6i_R>!l6c9bC_Ult_!Il%l3 z(Nac_ZNHnKjUSK13QsbRp_7x}aC{;Ni-LXEoZRK-o4ybi6487aoG40;8CU(}Yq&_X zmc8GTcpMG&W1zDADJCci45lQWEhmrU7&1+Lk%rNmTOMj&e^1R6!{!$my6sGcymfXp z&;o6JnjjP`x4yx7ds#TCDdgi^%kml}1`3fLP#-c$ofAU!pm6h|xZ6tzWM+frGq;s3 z74*{DfE}u`HFGvc0?Apba!IIYu#3Mk+XnegUaIvO$(Mre?dizfO{)M z$d}odzs?6A{aQK%U9FcbliJOYi?4-_#$ulSJWipaetlZ1nr^s?phs11;IjHW&O{`Js3wo@!5|2}KX&yi;| zQx++$CQJs;iT$;ADO)Z`Uh7YryxX*l)R>wK?IF)b|@Yu}VvH$Hq$r zX9l@QLHXBlnpf(&{wDf3t(ClzI@P@qggAZ8if#YK0^sH??hdTt*`koua4DfK9ytX4NFMydL7ohbnPC`j9V2)SCWo>lvUeN}75%=#)&Kc3E$+rHEo zUj0$TD!=kYdkgQh+Un2#wGNxJ>YuQ+@c8M(svc1_4~EqC7ln1Sqt5tPUhE9=pZWrw zib(U!8GRhniVay0FuEt>-jJ4?PVE@YkfQj_JSAKl`Dv~g?vq-8?!~)4_5u_J_mDVS zo%1x3%`lIE7wmRRH(MN}vlef_*vEqqg(QqA=L59v)DYU(?;CvJg*rv{`lLn@s~aNL zaBlw0KFmJQuePIqQx)Kt+`&^)PrP@^IGbT2+OjV^m>Jf!PP?Fzg5^RyH&s-<$i;GvGG%}q)=@52GGC95;sZ@Jm!L{r!z{_9Pc*+xeqj=^wm$C z=3$K3{_N{{Bd{Yc5wU&y4g1A&rvS1GQYC)B?x&#|2cpR}0Srq4f9KGvQA0ew*T@^b z`!1z|0zHZ>s);Ft#m0TclZFB^vgdtQ%MMR{3t1p&F~wag`bDnP^FddPA@9e|S3ZuO zW=c7QMYQDX3YJ^67}fs5Ym;tM$8R8PwRFerg8hayk_8q>)Yzp%tv)u49kz~96SL8= z9U?bn=%!LxoXPU+WUkqyZ^QA(um%WzJ4r9~iDyz_xKU}SQb^?U+p~xpn3B49uKOIf zliNfJ1*tcMC+tw z^&QBrI^a#lX09t!P(&|z=^4g^UwL+?Wm!seu=(+zhab}+m_uL#LqQnZg3gvBYV;#ao%8gkaN1)!B3m&XeDRq1EA6w;?8zkHy2`(7Ss$uZUwcp>myKWKZ- zpFLC*W?nCMxQQQ>6&^$3Leqru-CpAn*nP?7F|NO7q9iYc8;sTp$BSL>Swk&^ zFxkJ-o_>OkKJ3PH8DF=s?ehCO&jVFAH-735FMcs<>)h_P#HEZe07JL4r?2nqFdOHW z)+#|eL``j#@nKwcEqF1froVqj4VmHu^#Kn`E3K(!sbf;+*>>&9_l>-cXN@H)*c%-7 z{g^*%h3rK*ko_1TqOF0PfID7B0!MNm!wk-M{BlW{%%SR}#%*E%9VRt(Pl5TJ>H}5f zh~p%>3+3jFgTP|RFyOEeq`TPh1%5xZU)rY|TgDE-F{1~{t7$)nl!3>Q`*(#l&M4=? zhaxeY6$SR^?cOMIUgq87mou}K)d?wO$2;VizK|?Ne-WVhpb#U2@*zoR1d5WN zlt)4Xj3uMn@|l!!!+iloVuEf9IOFA{?elt$k?BS6W>Uvy@$nBw1Gx(#WX2gwQ!D7{ z5=35c1@WKT`2p(XfN!(pG$-2RueXWeq%I!*F0VLTMSeQ-)(reS1(8QcT+iy_o4kR% z+l%f1J?Ssp21oqt?` z(=N6GO;QAhD-x{Y=)>Y%3L~dn1hs8h8cO913uo1blE-SQKrCtZR1e1WX&dW zMw&%oXS~Fs_EZjVEBm4RE|J4RGankN=g)?IX!OiSlmy-MQxo&!+!cPY%zJbGBQSiW zq@3DG2s>HSMKHbohlz^yt6ymqRP(YTfA+0d8}F8wP>d@+-ua7;+mpSjG7k-_O7`H; z(=%H=Sh$eLmuF8bLywR2Aqf`s+6CR(+qxKHGg??4wqo{a2!yC5uWbCGDl+SYxMTeVreOia{$xT?Rp=YqeL zR!U{KqN=vay7epZ)QTdUMPbgbhc!q(LC4*XKV(tuQuW3)Fz z%qIK$)wVMN$&~@#&CFDZMy`u`hyUWATXKUupUTQ46LC=IBYIkT^dk1V=ggm$astzV zXvkvfjwWqHUh29&R%{OULUkSaurov!XWoFq(QSGI;Q@AInF}FK+>{j?6bho*#}wty z4y87H4o(yfJ~=BE9vF}R;a8(#XciqVh%E_alf`brHu^rR0tJ_g(RUs%wD-(P|9#wq z#T?Hmd$0X#cQw%{Chn#(gK6dlbJABGU+oxI(zx`bIoX_wlYUz4cTimBELmJ8impjy zZBWbB+_WLZJv6)GhhX*ioXo7B1f!HlRu$fseIYnF@#UWo*4ps%4ZF0rlE{VJEF)Da zS=NXkq54+q5_J<);?jdlJe&CUvVd2nJY~?nnKI-D0ZEd;(ak!-rdP&R-O^}>Rt*Hj zb==hRShK5H^ccT(qq<#PgVbikw~*79EOsgQ z@^9+GjIzZ|cCjCMOe2ga$-pbU@}w?{ljb!@3f*#`N6%E^5HR0K%%|bbgVVX&zU=Qh zzZ=$YDM3y=GF$Fo^j??O`PFHiJ!|CNDGbDOWbEy#>+oJkq*%nsyfhoOKs$`Gt1;Q) z1`k*eY+wYDvzv{bSR`qt2A;T(hixHpd(53-K#K$;BtKHzZT+d9p97BJ(YRFG5+5xy zzOiin@OYKeIM8KK4x_%?aX)idXLkFo0j{i35^3p_iCoygIiUC)FMq;a{&p=?rxO@2 zDmE0g@YV5#g9*?7{{dIov%7;_fF z-eGq}e|ITtx1={$e;kvquCKlpx7K6I-e8(q*UkSb&{)LpO0C1n!b7EJ!C~{^%@$H- zmTrR;MAc3kkld^=I^dVJ0=CbpXa4@Uo6$x~1jPY~y?ul8CtI<)>r8B$PDC^#m9>!m zXpDiuxHCFi#_MYC^dKB?0snf4+`$>WLy{sX$*M?VTdBGZx}A1KkM7E&YwFvm_BdlF znoLdZ6rrDd5b2&NBR*0tPj(9p^**be7=WK-tT%9BcSDM_ zBsr&GFzaLX+$}HWIezKv1XRH+3SdWwc_GIfhw-5MZy{>N05M*WJrDel?t26~IuBm% z4>6aRj=|j*y&p%v&3&udh@zCbimVLseLK>Y}U$5)zi)B}t zB9uVHmV6gBqxXAA^qpYLKpJ&PA{@mQvm56|gRk+5CKb@CPrr z{|pPW%kt^WqTOj?43Va%l{30k1NW`AIuiDOK)R`bSS>Gi2C`A6sMLYgPE`a^1Gl$g zl*I212trjJmCFzAZmwV5yz6?DD*YY;2g22_hlB$tND4~Gh?h7re1o$IgV+k%T;jdShFuwqxMy!cn z=Ir=(Xz$T`oRw|f_`g-X*&$aj752B+%pmUnQ07RkMbSvI~WWNFC&KzvL zoDZcH-#8`pGkkWbe;eK2WtAeGJZ)fA%k}j1G13C+xg&eAJ2j-HwH);%PY(wS0Z`&a zU2r6%7qb_JqrVzhZy^>2*fOsI~o2mGuqgjTU4Oi#S zgh4)*kz&Y2>}~YLq=yvm9p#69*;Q2Co#$ITJ?=71jMai+*6IBdSctZ*4JIyphGoe> zHO!nngk4_a`6)yg~6rQ(;=d~-jE@@7#-kdxK;1Hpy0&8eZEfeRRsvAG^y6Pvf z-2<_#rlk*KyCnG>_9q~khF24i{A>I1(OpjZmw4$tNhk&`M8ol!0`$v^hYiT=WoZ0S zM>tcDm!hYs(G0}Zq$r?*unnwTkP>@+DnowekLG<}gUS*%!$rviK+Oo3N&q$B{{A5L zL4+p)CfQ3<14}LWF(UC!7wk&hy3Em!n7SqRp{GwfiqE06y(^6gj#ofyWA_Ey(}-7A zv(1j^+1ppF(8NbtRiJT&p`*7hFCeXvqqQ;b)HJN$)TnHX#|UmJB*!YVm&yN0YnnI1 zvB+w>h=5^k@mhiQg}WBh+r#&@lX7yJ`BUvrtpQU85tx8VMQ zS1fK$%4t1u-R>&XJ((W$kS|D*x0{Y5$!gbgd=Ieei*4d?0q}s=qK7y^S>F7qKZmpV zYlW$gP74MtUT|1?kUmN%jX$-i&J=QK0ckeyxZGtsak==<=&Ks%r=vvh)=)>0&eKhtfIcWp5J>EeKDm#- z5PjiXIHa8cX%snzO@-VA*Ve;Y(kZ@=p79q|wNj(>x~e<5_%+(n6$*u9EcGS2uj0c8 z9%C2wt}oGbZA5s&G>lPxmI6PBe03R11<`%`y(@>lm&9b}evv}~?1md{EMQl%TDEDngg!|St{D?fn^NI1-h*Y~kUwMY;}*Y0U*?k*)5_iYGR-^kmM+7hF-Blgg zrr`C)y~bd2Hyklbg;;8X+u3MzvIOohL3RC2lFOF7oa|dGqp1e^$-foDi&9T;4nezrx&5 zx|mXZcX|DQ8xus(0??30f1;s~N8g(hrP~4de&}JK353vdav*@7>_>J7j z0xPqXdP*ql|3s80V3e=I+mXSC-ETip#NpyD{in>l-AQSGjAl)$B2;^C5BjM5Xe5V5 z8ChAKK_9{4?EVe9L9{9J7xl{aRK1br=Fls*j-*=oZYT*QC@lnQ1ER`&+BzcBmn?T9 zfwXZqMpC#*=dN7L3$c~~gC9PZ)8bdfkPg0szn#z91_wx(zXIZ&$g>nJW?dS zhR#cC-@X{(#Q`rwB$Q6occ%{Mtt*N5Vo+Gb;4%@yUi#v?Xjp5Gf}2~$sj-9FFknu$ z>T8t!LrzGZYM!lm;fm5n^;(aM5xr zD%N+VO9jF0DZiSlU9aN8BQ0d;lesy-MnZ`~$LSguLdZr%89>(|efs#GtwuVq@d-o2 z$6#|21Os9OhlF>4q=DSn=2~zu0Mx1L5`|-ZMfm~=CAnrLPE56dG96w{c^O9s3ss$# zJ3akq_V78^0it0^V>u#sFdswQ4#)1La{kRhIO?W5Ty;4Pp(XssRt;<<@$_`m;U6YJ5=NU8 z%`0vs6kwn>4Pw|zyO$Xd0~MIGAF7+~KkEKN8kmN&7PP2tS3|E6HxQb_VF*2-)qI0CP-V|aY0d^BtSbYJm1c{t@XxR$?^A&6YKu0c+(RhjEUpfEOw3)-X`VEP= zk1$tFg){qKS-G5x-w4V3zTm>}Uvg;z(Br@8@l*MJ5b19_Ni5IbtA05wvJ(<-L!=g& z103Z!38F7^5IscliVIkKZ6rTs$&~Ijd`~^Vnu#OUnQgs#t*AfYJ&(RadETK=8(rM3 z&b^y{Es^v;$s{6-`Y#LSe>@ET*%FjpwiBLVdzs<2QqPBw%^d|`dG$fq$tBrwpl{vaNIoN*R(@1gCb@7A0 zdG!*LCDYBdA~T-w`>c_6TzG+xk3&|Z_ayS|eD_{-Gk+;mKg*of1ukoeV5jZhJ`(qGMZ8g4Lx0pLK)=96T??@0J!WB2`p0AIJ znbj@0%o1te(S%-Fo?g{DFy%&yT+FHH58(zdrC%#z;mzlEL&7i%@VQP6SddmRWq|5a znvP6O*NRARQI?v=*+}HsluF;0JYLh()rS-q(W82Ko&tQbKADmn;m;vg`OqZ#ptpVe z+Ybuv>CGOGT^4CDiP#`|JT%G|b=_|hQ!s{`{wnaCQk+vAlGLzbl~mqdDbnHzqjYpf z+&dPMqhWyxfWkXsO%^}3tg~gkjuNG5tEC>x&+WS34g39G#@Sddo>j<|J63|)2D<*j9P@W3exKh-8a*OJ0Mnph|FZ+rg(wdy}RW&VGe zhyO*Rr>wW<9$JKc9k;X{;*WVDmwPkd?xAF^HPWdLl1VrQfP|)O&dfN00?+@mDKU2L zqdqivp`*<}C9vJmf|%=~{+{#5KyMKY)5w};lHzROf}tas)1jWQKCik(-_Hzz^PkW1 zmY-MK+CmJM#d@Y5Th)%rldkghA(JvCEHaab;|A}FBZTK{^^9j5z4uUyv=bQW8o%vi|)1~4}d(@lT z9~%_+)R!LO2?x=hvIblSSpW6{!#>woJe*dhXse5%1uMe0>!^ULFOM4WS$y&zz6hkr zPlPYY3`%mm;ufPljfV5J44EnC^s%_O7+7YkE*5_wAiU80`@aD7!YDg0kj!cF3nBxm zgE7A2{*2T`Wl5>OW93I0rGR@Jh~3J)C~Q2mND*ZAvm~*lelP%tmMw(mA3@o*yD~Ro zo0t+tGSV;{CRxhS=1@5lxL~gGqa>xO=L0MN5j3Y5)_iU-Q}%SXlRVh>PVesc*)_j{ z!PSrBU*?r1LtFPWj#G~^*2F9X=9~{L-tD)>w+z&?966q20PJ!9xA|Xz;b&A%r|7{=s*0M@&u0`A5fl__6cG^=k>~Umign|2UIfMZU&Oz`TEzMpeu+Hee?7yJ z^>gX}$o~<3{Y(5e{CYmDe?6y-*3V@c8?76k)29FwlwTsxAo9!l8UCaHdJc`x$FIic zbm`YpFH^?z5S`t`f2Ic(^BKpGwR=)p3^7kK5y>lXMN-PP0t+#g)HOW z_(IMTl!$`z4nFIz5k3u}93nbU|fJDQM) zi1w?_3_Y^fS+cZbIi9=mJD}0<^8*r8kXJrx6@SNJUWM<*U-Ff)u}uJO4=P5$m81Ha zvtLs;M*4_PBUq;hN^*h_BB=a=6ASw7P@}}Z;Z_}jLSn%oW$2WPL zrqC|y_m4MEQ={?xCl>M*M3$KqB}sVOrCv}9QjcNToYO|mvTT6O9et+KH&AeHcDws(1aHDZ`lEp)*pUy zR>xsDscDuph)Xl|hI(1ayyLbNd{^B%;&-896sEbGyOUy#cM|OPSaJ?*>0Ul5_$%_t zhjG5|jUwYlHA~@)8YAZTNPS3>;v|-KR+50mK8IE=>Tct@Uw#*inlx zsAz!}^3+__T4@9r*o#gW%U*o*;}6P<()?Dqj)5h%%=o0nBb*gX`{^$YPrJLEcLCyV6k1jT*(-qf{mrIxMw+KiU%aNbXK zQ8XrjdZK(n6zI?*>J74kE5)xQ!BHx z$&yf-wdY8Dnse9lbGZ0ANb|)a^kcZ%C{t;`ZFi&Kr}BW)&CB5O4yY?ne<#r};k-X{ za@O;>NtGC(E>RLD#kG7|E3@$g!b%aj(V$RAJ%mps%f>X2V%Df1hx$`^lY}9M6Ia# z^Mu}6v1i8y-q4@DJyCglS)YQlMTlPm-J}g04N6YaMfILm-0WuyG+O-D>rT$9X`j;N zbyeIerUxoB$3it&Hu&yJM?-PXhKoXJm4NCEoCs)HeqVmcD~{r}@Tcproy4z?=ru~( z^9LqeaMBrqtnQ04fm4-B{3Na{>%S|tO<%* zz$awbt_r8v(~3z@N!kaLuFv8?SilNPz~$G5;1?*IrEd@i*Lg>yj0oc&u217t_667| zu4)Qv(56v;*k4HiqQ2#>QGE&pg|q>!(_t05(-F&+j>pD`^IS@9t!aGZpNISFEby!x zCCXY}dRJlPH8aXTi|7FfC^dQ=#9jGP&}2@Y9A(YCQnV&p zW;Hy-zvcytX{dj04%0GLpjwCZ3zTMF?d#X5FIBrYkRO>8NKyWwR-4W-c{-~@Y@VI) zY}#UAe|mF;@Vzo4RgomY9asr}1$@Dhn=W~?`SaGgJpc{(A~i=JDfV&HAMiM!vQqFz zbvePcQ!v;}o;ER`tWaB!>Gs}`wovU-m z{k&i^&y5~xHkV2Qo_a2rNwc4hxxl^FctBu|4@cG1njfdvVBEv zIGUr=L`v?$xqd?!`~V0dP7+5$S9bey(7^UX3~a+KobalVHGSkV|B1uJ z>Hz8JYo3L#rk@hT^E)cjTp0nY5_~U@50|`=ByrCbF))%CCqQP=1Uwrg>FUC|{>dAG zu=A1;kU7S>y{PJ6`#&a{XEDL2sy@^5FQ8~O8-fFUnaNp%?|)NH#_GBA${ZJ!y=yJU zy?zm*HJ{M=t6ms|K=_ui9xo}5LnKtnjpB{epe+vEWGeIp-@R(R2Cuu-iWo{9$fUWB zRV->fUetbX!-^gj>i?B=j>M5ci|Msu`#8oZzoZ`S?ZS-~L)Wb&M8z;L^_-QA2<`f; zo*CRrO&-V3qTkpfDoBPJ_P6>>gmG-b>piFHjxaWm-sfRNKhyr&mCK_s6uRXaQ$xYJ zx$F^kL;ZvO?&8)w7L^pU84#9LNW=7AT`sH0nk@5|2w&-)JM81hlOFw!bBq#};nz-T z1^Dz_Z-U}nhmIWPi3l3>V%OYt#-aKW-o)<8e|Wv;>SIgA{{5Qw`;Tsaf#y6(Z#9rL z{lh4xp;mLCaAs}tR!Z}pboCoGp8<58OXhO|Q}KriOhc2#tZ$j-C&4`%Q_Q6C*kj|? z(=(U5bjReT91l*IVhPr8>ajQ-)4sL5(=uX~_LZQo&m&DVQcPwOh6Se_`r$V8PRrB?^6({ZP-2`BZ>pT*R|108>9MANf!7ZnK*fI>5Hs3WVJBO z2&BLU@@r4*Io7%&=@Q$@c{L_B%3H1t<<`si$xzG6cVey1v_EPe$B!s|p=8vE)X9h3 z3426sUvk=gUb|m#ky~r%Tln4AcWsYstwboJ7wzP)jF(W9^Eh=5;9~h_He6m2hmJY$K25tlspx z{d&`dplMaDld2gwqH_G&fftQt;WwCm0kcRVt&DfUKq4TfT4!O|V5r((g2S8jnf#5| za$zDhJLO^Fe($qAaiB?O)Ri~ks$M?}3>D`6YC-E;Xcx7b+kBaVvo)K!=1`Eg1y$vi zOMHiU(X{5vJVl-uX)?!xL0wCX<#j>&zGZT)7yWk$#eU~6JKKPNh22Sz#tVNF)3U@j1gxbyOvwkX7+D9+*=hYxbZR#M6iRI(S zR53SEui-QFH1PYQPZs2U zqIX$7z>Kqr_D)Hxb%I%9DBf|aESD#}OHynF&lndVhkRWi9FY+a)~f6o4D!sB4HD_l zu4J5+%2&yL>>`du@fKCBHeiR`PA&LRDmEu~ZVTE7^y<7VZut7+@`fd3=K>;d&=gsx`YAOq~dKBAIJKjN5HeD8ytSxh~XXbKz$<+44QMvZB3%?lY%x4?t98U<=YdR!d#Cptg0W7J=pnVKTy7|l9&iyK+ zF`Bb;r{TK?OW@{+r)WDrprYWzjWy8b`+ZipCvD^^OtmC&kZbZS7IsdBN}*&eZuF)( zp+5)*K*PCutrHvccs5;lXODz$zQVmP@qAs>^BIuR;63#HwrKa#{J`FQ4#$gij-Ti| z{m<3%oyDX%8qe-33rz*mm5Au0G!KQ#=4wuWq zlMrQR@-`LQhgS%3s=2c!tu_X=ZVc}MHns+SMv`d!Bd(UXfojbuM1`4aN`98dk1n;H z!mD}n#t*JCqN*KVnAQ}mrTwL^d%z>SQ|BVBLwXJKF-xz*!1`0#ms*15B~?8p12jc! z=uO(^YIHsU{&97)V)ruF%ZBB2^_e#k|~tfNUl0RnW0nK zka{-km=Rqc++`d`=k#T6?IQ_Ks-B)_B&@bl&ge0kJRSN5k_frn*;Jd{Uisvr#)D`H zUMmFTon9lJ&*BAtU9FMQ?24=jSuIs?n}X6~ z?N`+f&*(Gaiaz2TmpE&To0d)$cFv?^{F=Ys9X-i0%sWYdAwA>ZHiW6Vk#ok~HTN%m zk8L-CozBZ%?|t^j*;4CtUSObIx=Lv{exUzia=E~w*k}~uhd{xDgmVQFo z0#YHq;=0JTb9T2W*KCPiX8A4!~mOyxuujzOa2F$^J5EEM^Ms*y5Fv@lG&l%pU8$D@kCM( z`RZRD9x5iQFBZ<#%@8{O?L9rI)&0(z8A$zm*BhHe@Z^=Myp&GQY-KO6zJ8&bO~Kp! zhv99Xy9JihQ(0x`UouuL$mwonf>-++zefYsNTCQ=ySPwLh`SjGgp4R-HQ!QHw%%<0 zi2>;`;k#WD`N92&n9_{MCF9BWp!NJx9Y4aDRQX0T&|RdT4Lt*o&!cBYZEcp;n#H`? z1+wYVR^FWr@|P$>MnoH2L`BCe@^`o%ELoLIqI;O8ph2%o&kg2gYY&$8sK?PWFcKw4(=t#_cx0mi}zn;Cz5@Av6 z$Cy%xMP#@(%MqahB|N;#0QsFy(tb4&O=R)tlFf9oVOUWFkxBN5hbI2mRaBUjQ{ zc1kZ;c_Ycq6HaV5G$#*x{=z*p1%%d?+hV;o8>hOhyfa0@53xQxocd%U1}tBIR%AUe zH50Y1Z&jcm8_W6%dMUn_Z)?)}r+p~q1pupm+@5tShnZ41fOF_u(HD$k ztjcVS1DpYpsP10RHE&69#~nDwWF}n_1xliOZLt9OiaK6VEs3~(;_a!rO_^t7hh1Ev zQVVV)*`%44k8-nVMn>h3JPmq57y{Upi#Q~F^XI#5``^1`Tf~)F#{JKAsVuNlCrzQ> zHFbPh00y4;Sz3*DXdAgx3R-e5MJ@wos@g-@;a`A|u(O4s{r2gxYKmIbw~dc?R}Xkx z`WNeaOQdc0`xG^`G+F_`Ia$n2~gI*ZvdC>GH~m?k|>t@$!cLQUPvW zn5E?scERzm&JFcsB#YSdrtlHW_0VSh_1YLnbWq8!T?&4)#JDcaq^#(?qf3M{j`pC*yYOZt?tIore^>b*4KRmV8K)t9_zR7@7mnXQcjyf`*ODbgA8 z`~!FuA4q|Hul_K;!OpdKE6`Yjz=6qsJJM6uHQ{-5)2Zz>eOgr8zfJq;=fz*1?9Hx- zL843P?qY7M$XJo&%%To{4K|UFf0i4NP&N~>7gwKk%!&6)npc@?@(V4RxZqsiWbb5e zH6@8i;`$Cd1GaBu zfuKiuoztYAKR8Wha+&%#@K=zQtrr|d_mF&8lp-*xK*<&5=G=Q?TGmCd&$XlsZqfJ;_1pQh?!+i(?S%ns}7dSJJoJvcye>_?~m z5iM`%227pKp!bb`VZRMF)!t-&fWlzmP6Vn!)&!n5`2uU6>3__}dUvQWTeuACk}(4;f$p6lj=s|2`Y?`3n&^iN#i+z) z%y!6Om&j_1YuKbt1DNo*pF<#(vq0dG2q{wpWePE!rSWh2K}J+De=G26py?p@L%cVe z=3c10!p8bdVN}p0z`1E`lA$Qky=MEc3l|RhOnqO+da90|fIJ}T?Cnga_;EFq-j;Mv zX@m;h*IqP$SANi*;h2<(6*dse@S=^puBk8h%jLl#5J>G4EHDS0AKYt_t1vRJ9)Pi) zGF(G@@f|g1=PN8@QjWP=T0hRTFEZ)+Mrj;+MmD>UZq!ONjMSQ<5Q4kMH><7-
WA@?!Qd(4^tT^bVWw%HPwQ z-+Q8gsoqipL!qLA1>^N9o!~QZHLg!?VPi2pcIfQnm#=6Bx-4F_c}2Lp8~KWQ0LTnw zc%LNjW&IJowgbJhp_j^*q~ijd55N2SOlxDzbG`ub8#6IKdGC_-oKunP4>*=FbK;Q5 zsnH2R7Nm6=r>Zt)MkStPV$S$PkO%Y!HBSY2c?qq*)H^|WRp~N=a(DRdJmHn=su7L= zBJ*&Hfv8U&6Q|4x+phHx`R?TQukzX&!d8sBUV)3(Hj$EC2O@Affd5C324>0}u<%?d zq`}`52v(z;pS23^(jza*^!xy5AUc8cr>Tebd<}w(TIPM>ZJUj7DK;E2o6b>sY~R!M zAFs*Q{&{JTH#*wCnz)$Bx-yWOsdJ-)1Bv<4(x%X)qJ|Q)RH>EX@#tMG`>XwrX9TY< zLotR-y}=z$5*E4ReO*BgDK*=PsEz_)n%3w^i2S#OWXuBiY>fCsw zvg`;S??-{FYfS=rLQFE~#lsi-4hM9=%W2c6LDuQDYOq@)I~Km3%Li++YX)kT6TK^) z$8Lfj4uJijTmdc-fE&cr(+`lI_2vKHvN6n};suWn5tn6#MoN8#0K= zW#=+{vMU`(IcWG!_Eb?O|_vI2$7IgiC1xNPZR*nPGB>b6( z=gt|iy+)cfZ9z+adc1Y#SChBLBU`EF@ZyT72Ld&IR9~FR-Wu>BmG$hcBF+Ul&_st7 z8y0^cy=-w4&@-f|da5M#?$y;P7u0as|1Oz5)IH{q7T^dx6t(ehSN5S-tyOvP*tN1w zG*tM#v}%V`dgZp&o(ggvVeYBTXZyywhzr{-g(Z06{>g66&tbUr|M3|3=>cB{^N=n4ukA-sf&AY1h& zwdw8D%V^01qn(vfPAZM%e!nBI=ZL)SDeq{s@nRhNT1L{~3L0x21f0#)hDR&S47{` zjC|Q@vPBn^!$4e5zHi7r{BRwbfl>buRZi}42wf!zLM;QRbPW3^!e!2w>{_y5SXw{~#w z0ZTxWH&ttDycznEsx^m`i$ih-5R;0R}%a^ttig6%$|HxmRAx#o=gd+BdA|TZ`MSIDV{MGvp z*(+Q6sNhbg)dv~m-Dc5uuC4HUtI%UucFV}A)*r%z?htg@dj)14^Qv!O6h9f+wHP1# zA=Rc#(YRt>lIle|-6~YcveOpYNnwfcGj%vOWJ8|yt9-_=v)lIv+6Q|%+TUFm@r>hp zw)VL6&D@cPY!X+#YM1;(Q*Sjp-ND_;Z*PjflBQmJ8`3RCfH`ujk#Z5=G%G9}`ng28 zsDZRkNSnyLhT^l&=1I*(;R!*@QL8^ASb&CjV^=9Z(P>xj}16)=BUJ@^c9S!rX-Bw%c5j;(Ys{CWjsFdO|VJalNVL8gWf=X=I-B< z&7NFQG#3?E1nM{e(0@z)y(bz5;62B8Bh)k6Z6wB!;);TUCNKXLaqL=zOM$|2{DPAt|(ZCG2U&=Ar7+y8i9E+4$SmDggka{Q<66nf4fL`|7n zx|NnxK9sDJYcPy|fzb>t%%_rw80HZ=g!w99c1FqcfNX&4G*;zHVBuHw{aL!5uU0i~KPRw&2C@9JZ~2@KhBL3B8Z%ObGjQC+o7#!! zg5GACN0GNd(PRz8Am%`R&|r=}52Jvi0Q<)|@!_>U{x#D)kNYmg@+Q3$WXB;v`iI7Q zzuxzpCksCQ)8qC|p3}#6Nf{Y&wA{ycT|twdxHaO2hrCyDtPt?1>Bgr{Sw=ednfAd- z{5lUMl*3h+^MAaLyHxwgA6g!{KJG9Jn+53inYg!z9j%NUR#SzCVoOkc93Qv2shPaV zH2YanSd`*_E;!+o`21E>4V#=N6cF4~I7Dui=D%)&

The 5th row of this data.frame contains information on @@ -1978,7 +1982,7 @@

Nicholas Clark (Nicholas Clark (plot(log(MCMCvis::MCMCchains(mod3$model_output, 'twdis')), +
plot(log(MCMCvis::MCMCchains(mod3$model_output, 'phi')),
      log(MCMCvis::MCMCchains(mod3$model_output, 'tau')),
-     ylab = 'log(tau)', xlab = 'log(twdis)', pch = 16, col = "#8F272740")
-

+ ylab = 'log(tau)', xlab = 'log(phi)', pch = 16, col = "#8F272740") +

Our rootogram has not improved much with the addition of the latent trend

ppc(mod3, type = 'rootogram')
-

+

But the residual distributions look better

plot(mod3, type = 'residuals')
-

+

We can also demonstrate another feature of mvgam, which is the ability to use Hamiltonian Monte Carlo for parameter estimation via the software Stan (using the rstan or @@ -2116,56 +2120,56 @@

Nicholas Clark (summary(mod4) 
## y ~ te(mintemp, lag, k = c(6, 6)) + te(precip, lag, k = c(6, 
 ##     6))
@@ -2175,85 +2179,85 @@ 
Nicholas Clark (Nicholas Clark (plot_mvgam_fc(mod4, series = 1, newdata = data_test, ylim = c(0, 100))
## Out of sample DRPS:
-
## [1] 90.56627
+
## [1] 89.19212
##
-

+

The trend now evolves smoothly via an infinite dimensional Gaussian Process

plot_mvgam_trend(mod4, series = 1)
-

+

Traceplots of smooth penalties indicate good mixing and convergence of the four MCMC chains

rstan::stan_trace(mod4$model_output, pars = 'rho')
-

+

We can also create quick plots of the estimated smooth tensor product interactions for the distributed lag terms, which basically follow mgcv’s two-dimensional plotting utility but uses the mvgam’s estimated coefficients

plot_mvgam_smooth(mod4, series = 1, smooth = 1)
-

+

plot_mvgam_smooth(mod4, series = 1, smooth = 2)
-

+

If you are like me then you’ll find these plots rather difficult to interpret! The more intense yellow/white colours indicate higher predicted values, with the deeper red colours representing lower @@ -2356,7 +2360,7 @@

Nicholas Clark (

+

This plot demonstrates how the effect of mintemp is expected to change over different exposure lags, with the 3 - 5 month lags showing more of a cyclic seasonal pattern (catches expected to @@ -2410,19 +2414,19 @@

Nicholas Clark (

+

All of the usual functions in mvgam can also be used for list data objects and for models fitted with Stan, so long as they contain the necessary fields series, season and year. For example, posterior retrodictive checks for the in-sample training period:

ppc(mod4, series = 1, type = 'cdf')
-

+

and predictive checks for the out of sample forecast period (which demonstrates how the model tends to overpredict for the forecast period in this particular example):

ppc(mod4, newdata = data_test, series = 1, type = 'cdf')
-

+

Logical next steps for interrogating this model would be to trial different trend types (i.e. random walk), replace the distributed lag function for precip with a standard smooth function (that diff --git a/manuscript_analysis/Case studies/rsconnect/documents/mvgam_case_study1.Rmd/rpubs.com/rpubs/Publish Document.dcf b/manuscript_analysis/Case studies/rsconnect/documents/mvgam_case_study1.Rmd/rpubs.com/rpubs/Publish Document.dcf index 0581f3da..3b4b7e99 100644 --- a/manuscript_analysis/Case studies/rsconnect/documents/mvgam_case_study1.Rmd/rpubs.com/rpubs/Publish Document.dcf +++ b/manuscript_analysis/Case studies/rsconnect/documents/mvgam_case_study1.Rmd/rpubs.com/rpubs/Publish Document.dcf @@ -7,5 +7,5 @@ hostUrl: rpubs.com appId: https://api.rpubs.com/api/v1/document/949541/2ff60f7d573e416282d4c8ba93051852 bundleId: https://api.rpubs.com/api/v1/document/949541/2ff60f7d573e416282d4c8ba93051852 url: http://rpubs.com/NickClark47/949541 -when: 1678506655.68476 -lastSyncTime: 1678506655.68477 +when: 1679992782.51067 +lastSyncTime: 1679992782.51068 diff --git a/manuscript_analysis/Case studies/rsconnect/documents/mvgam_case_study3.Rmd/rpubs.com/rpubs/Publish Document.dcf b/manuscript_analysis/Case studies/rsconnect/documents/mvgam_case_study3.Rmd/rpubs.com/rpubs/Publish Document.dcf index 1a6f4a28..1ca6b513 100644 --- a/manuscript_analysis/Case studies/rsconnect/documents/mvgam_case_study3.Rmd/rpubs.com/rpubs/Publish Document.dcf +++ b/manuscript_analysis/Case studies/rsconnect/documents/mvgam_case_study3.Rmd/rpubs.com/rpubs/Publish Document.dcf @@ -7,5 +7,5 @@ hostUrl: rpubs.com appId: https://api.rpubs.com/api/v1/document/863866/d3c20f81456d4001bbedd7b7724e43b3 bundleId: https://api.rpubs.com/api/v1/document/863866/d3c20f81456d4001bbedd7b7724e43b3 url: http://rpubs.com/NickClark47/863866 -when: 1679287676.5247 -lastSyncTime: 1679287676.52471 +when: 1679996632.30156 +lastSyncTime: 1679996632.30157

niR59eSa76EbMY{?!29@ z%iuyH9*g3<`ADsEE|kk(?7Mt2&w(uJYI=WG@w!#M5!@j_E+&{%Ovn$)^dA!+3okBq zp8HDOGV0jr2B1Hb{&ns}W2=sYyRcx3ZM%Ss4r?8>t% zzrzYyGPC$tN^Q@mHf1uvvUn|6U`&8dRDYIp&(0FqVS|XFlE_#}$N8#dccfOYx3bS= zA5Q;c8vn;p46)NG;dSGsO4%}AitCBaldl?{{br^)gO_l5fTi?0t8}+X8`+SuesYNx zmRSMoP{B<_9n8!Pm5?ZCfA~)sMF4}N8cty)aVSyNm+DCtHB7EQT#uX9nYf4meeN1VGnNmus&xp)m}$))sm*X+xIxq_rCeu-e@{@;yXNpH2F zP|gi!~-kO!=xLG zfGeZ!ygY){)J(E}3L!^t|5>r%ukjT%xGa!okATC~mBWtsRMNnQQF&iM*k^k-jgUvl z(bX+AAIRR4*{wS)HaiT)*lzO38;|Akpo=>YRdJ>A0i@!FqC`ctoo}|ZaOrul*`c(j zDB`&MdUFQ4ceNV9;q3Z1S#6CFBx?&AOo`zr=JKSzCG?_A-|5)R;W#_A)o zPU=Os*-2iHSSnKN@&eXYPFdqtJB}|@5R&(nW{xsiTo$xV_03XgOFU4lP(X)y?7Hcz z`z-@nVHLXfujzi%6_Sra+Q}Dm<7=R+4Mul{8}+NROK%%%L(w-tmp`>jkJZLRYrFmHzxzg~NGoibL&Ai# zV)-c+zj^Jd=vg^q`(7ra1bY>FE9;!AP3zfOTFxu-6vWmUImp#P_Cu|17Wq-YiXc9K z{-Mz8N(W~^#KR;u&Yb{6G*D_}H0G2J=}ESBQUoc5k=~?cbj=R1m}}|d4fS|y;L{9D zj+KuN%v;lO2gMv`Ob-@Qb1^O;MMV5EGwLYC7AUh+bh2{0Co32P>K!baplQ`UtO?8j znkQt6$%x=mA9#jS^dX8?u*~@9`mC}nJ8mp8W*{Cz!~Dr?S*Q-e$yHcM2o>DcH9vpH zi|Bu~)*?G8-$LXjOx5deqh|sn&XIp%TUfd4`8^ zg<1F(qNg7m8QFnBDLPW91+l%CN>6)KcZC+0)>^tRwiikE;qk7)KV+qML8le{C41{$E~Zjj(-fv zN(|EK0JG2iz6?RmJx?XYl3>ONbOV%8mPg-SsHdRcs0lUKoRfOCs#?DXMx{h_9pY!c zWXI$Z{L*j?B-W@*D5##7rcN8v5Ie9qygVkFH~hqS|JwjI|E9XulNdBE^2&~OE329g zR3Y=*t$sYHZ0%zAjP}BYw(e)cKVX?Bs{0zku%aZkK2s;}+)$TqTAD(o!;3w9O@P8q zUMiJW*SHXk3ReDGN43=R&;0Y=!RF4U1RAALZknObl2nG{S$N2TY~`0&Z}+F})Vq@(xQ{`t33wKMB{Wxh{2psxl$`~+dLW#P=;o<|{98BfUni5#{Yl%w zhuRMW0Fn|ipr+%52sL*5-jmZ0|NNhok5J< z)Fd(Wc^*<>ue}1a#4%_SVTkLIkYI9kOYe9qOpPpo+A-!L@FMdNBVV6sM_0%JRZ&BZ zV<5Z4nkAL?75Z=;JxWGFqY9G)Mf;hs>za>_r%&_UPV0=_%WC8xO;6aIR>?g`Wb80o$)`nmztl(INv)PmovGB;G35liepS zqJ#ME2NmR=dG-p2lxNGG_SH15gprjRgnrU54#dgLsP$G%^M82^#>x(_&teN8d=Zr` zcSqJVd$GcMNy&9mCNz~q!HA<@=mV^oi-AVSPzCbvyG){xBWFrACB|Dq z9ndTXUL&}|@w?u#JaQ+d#~~(L7H)r&V_UbLdyM28SDSiL@(#{(*TFU*5xV_)@nhrN z2OYQjO~P##Sva?dIEE!imv`M^rgo+ji!W% z>rvIJu0=kzzxyV{k4l0*O#TMvgrKuYVQKTJjRj1XT07vh{=4ZlP$PDUp*$vE2@i+- zLsKmL?HN*Jg&M!feJ|$@G20)#yR5udY$IQkJ&Tfu)X&K53oIxB6%vazKBmb`{>wH) zvtAZKC40{P>m45Oq5um=(LasgLZYKLk{*;O$SaEYcGksN_4eOqRR*Q%K|YgofDeeuPr0Z z)7(C{Uxm{gmh5rEC^z5UH4e$^dYi8VoE6y!njr2TR5CKENd8K+|K!i_IVb?1x%l!3 z1s}z#GTD-^P%7X1H{FYyh+IXpuI_upUvw1ju@5uTn#3MQs0A+jvye_T%@#rO^!m}= zU@GJ`XU{{bLeFpFN#cFbxi~Q%$t3c;n3`4j>BsyHOSE+xCg3gz(&%3-Y6VZ1unZdN z#U$vL_yjHuuI@D>G7ou|)Y#x6M8~b&!e7hv>Kv+a!C&U|2T9^zOtI5Jh9=MI|mpTJx(! zcxJZjl%XeazOqKBm@hHy$8(itsfhphWXE{7#Fw~BuRn=;KopnN^?*Eb9|0*%#w7n+(@r&YB(In!hy`tv`7Fg4n_m;_PF!%T~=uQUq z;%XJ=-~!E3;irt+obW-?eE;QcZqCyxhX=j(v)z$mp$k2A5NmMi7S75LXF`!iR780g zONbS5;m~YEXBg;Gutp)G)ZxgfbZBK2cM*mS3NIw%;;J8(SsVU%3vT^>QHGb15liaR z<2WK(vwU0FGo;OnDH?+vSf7bLFzS4t8Gw6;C6s_Wn_s1#DU=vtz@uy6XjOSKA>+Xx z|JJKP8X^BpA|r@m5CP!^_s~ohu_)3ILu=eHr~*}Sox@7Dp^D66*ECV%8-z2;hJPTt zBYDnIX9pBUv9_QNB_1%ufF!lNeZVYb_c3TJyx`%R@E6xfI$?@pSjQh3b7N)Z^UF`1 zJyKzBJ2VR+qHj7Y+>-Mh2pf`##gayys}IJjU@HQnZCHr+@$rG*Oaa?R+;H2e;W(dt zU8-{`CUsMjTt)FmGEw%6%e`IW_xUH6(o(h7gO}%UhFh=8(RGK!(^t)@{koi}l2Iz} zIqsZqy6{ET5!=}~P16gB>hs<|i!X}_3w#wwG)Oq)eu3D^wfkVio$0u&n;8^^R_f1w@=8J z-2iQ+Zwb&fm`T+#OH9jd7BR6wyY8l@8;)a6^Gl_uFu@bV#N*TZb2lz;yw&g|fv{@o z#@6~m*J@dVM?Df|wrV!3iLyU>w$a#sSItgjcQ(sZ=1p61??=2T#haoAVQVu#9A`#v z*LS$Oo3Le}s|G)5`Os_8K5aEu%!oRQ`G4@h#mo+l%2VH-w%4M8@;4dKn4%-xX_^i`h-nrr5lMQb8(j^)1V9IFkQTn+Yre4!MAg_W0y`q8&% zC8-KR&?FJb;k#FciC7?tEHWT!?&h2FgEiHA_DVZcXGMAu>DGp5JhYu#KTDr$iR`VA~V z95O|Jelko$?q9fGbYVS36zEHGvfd6#|FE4wFxm%`_%50v-B069&QZI``7k@^sa}l< z2<08N2gnHph-h)Ait(fQ^DmrS&=)m$KZTB!m#U%1(%!BZ$!(m*a-P;TS+TiJRnaQ; zUQ-OO)8)Shc6di^UEN@Zk23`xKen*;FqxAnLHjfoE)Ct^+?8cdYE$npcWbT*1PZek z^w{FYE^LLZ{qz^O_(7a&(QDn`-?L$)UBvIEgi#$SN_DE^lx(H)q2QEQhzGOCi^Ts! zmK`@g)6H8RWurHCMYQ6;Nm#ivd%`eAxq=A%M#m+!U;y2Uo1{592YZ(~Atkhz9DQKP z;7ehvxJ3c45k-R5il>!SO(LZ8P*!Tml{dAi>$_U>@02}UW{UQx4kEeLn7{to=Nq}F zS_^!sMRDDzxDOj#dWJISe5B3MKf9&HxBduczT=3TkP+ug|jco39sq7^+cMe-Y zwvM|xzTM%fmc%B#j^@!@H@CzP7+B5dD^(fMyf*!Wo!<1UlCyh|Z<$!@Oclp~M|Bra ze^v6pgvg#$k#+|cni%vR-^{kLEgG86P_~J7%*=CQ?ia;o(F*gFU)tUO;5vg+OvXwHgRXkJ-zE#coI@35)e@=}bO-HDHRT^@Hur|{lj~?ptb{GybXuy4)UBHD4WrOWG=AZT4vkCnZgrbi3=6l3ks@Xwc8k9QBW+1`@&w+L@Q z_@mjn7Fy+CzfNroy9|t=s@vU?1Zk@>bFn)LJy5{{Tz)mYUVgP&522iBGMX_RKr}fs zE#;+4RylL=t{QfqFk;+1A=udbuTfp*);0w$oy2>#hrHAVk5*vz|C8?M*|Lvk;eymB zgJ^wpeyQ#!L}%ZHANXu0+u(b$8My;Wh2Ok8-N9Z}J|%Ac;o8S8CMDlSE7>%Q7W6QJ z-CX4YyEBlwZX@s!t#uq*rFnAmXt9QRK7{`H-@WlHuiNl_zM{Ku-D2+W-|9a~qgD0_ zWG<9SYf`fs)Ou!mb1NtB!)Vn9F&|tN~fWIxlrYDtm{H9bv1u-sK!T zx9N`+-qg2{X58ff@}x&+G-LPo0b98)VX!r--M`dT{jh~+-y%N_Z*$i2e0+L;QPiG2 z%cH+pl>;#0XWfFI(#@B{561+2{{E=w%xSVG-=&ep+tc6pYReU{kB(d)lI*!V?)dMp zfb4ads)|QPr)0@*L)N*_D#v-%&l;L~nD>RJ%%IxP?SD}SYaZNA`aX2grWRfR-6_=w zJE_0FC?W>^*#aC4767D>HuLDUK0EW}m?A(`@9?e0l@$WLOtQdN$03cqj+F<-nN^e~ z&;JT|`uD#8wh4M4mSQ`oC(#pxZqiEX4GRAC4}4dx95{Pj;38{VnqhsCBo<_;I-2uP z==(HI!Q<@?*I8RcKN(w%A=@^;>KM8;aK{bc5jx=4^}Bu`Zm!rn<$Qhw%Cv0ltVAKT z$(*+0&vkK+m9HxKh*x0~dIiZ49pND7GEAyFjOy&iu~(ciSp$KGOLLgr&;ys*;cs}~ z$59*>PK?$XyVf_6M@J>^V6*p;l)@zMhI<@pn(}qQWwWHHTvF*V0}i!KMT+a}*Uwp* z#H4}EtHC7=^Hcr@(a0_^k}Bu%3S&9luQXy2W|+`Hz2Kk+6@5HPgXA;|eu-T$N$R2w zF4f^W`IH^5bIgTi{IXafe%*YO^0N;r5Q?^O@aNv=>i&yqH7lyOsD{C~e)iF`1Ap>2 zZnAr?Po~v8c~r_+v!dvzpizQkA|$3OHJKycQvx`5_eT}xa`-EO@WZ>n`LgF*1cV(@ zbIOfcM^S(fYhh8{HbVncRFKn4|Bg?lT`0|msQ(MS0RO-2YMIJ`6(A;4K+v+#n6Mf0hBsflCOwML+$&p-L*RM)s&94--B+*{q^ENUM_P7TNoykuzcQw} zk9u-hk}8GCbMK}f8vo@?$4xRFG!Y`uxMsatG)Z+xc}^DoiBdp&o@1Mt>G$9SYTRA^@Ba1ezMsFW&lXqd-hSThpKDe1{hsFYrzL(( zHtn9;7RUL_eF+Q4oO4g#*dJPPT>H-*t;X+uf-^1~gsf)yt{lX`;`jV(Zt>@%`fu0Y zH+-kxb-?Mibr@5*&(s@-LEc{e`9hEJI_tRCc3QuMD)`-1ggIgkq&_+JwPnA4qq>Cr zL?49$#xu)$wX8*Fd<1EjeqjB1?i>0^r2^B}{geYqbJpzU{(joSRzQMD{NRT3+&k`{ zxaiEca3@&O>hg-uUv+oe>2KJZTBe=aZp)ug;{r_cFKv4G-^KWS&&gj{pYY$nc{Q){ z53prc)ANcxr{3Q8bIZf(UDKI9*UkwDO9H!Vj^p*fU+E9_$Lvy?xt;lJ_kKqIe_sB0{`>Ef?C$i{o5kO} zZP%4+|8Cn}jz2Xwem%QAbrsVehX=AnAJ=_;fBg2d=6C127ybX@{rkuL!}Eb@^nLuJ zmycK9z89H%dw1MjzsLM-Ex^4Kx9t8s|Db+DPN1UwPXESlaT>o;|9^CrcYp61vrqoN z$+o{PKReFNY`d>9OZp^yPb@ctN+}*Bg3|)PiwgZbp3e<^7<}<|dT{W@{hH5NQ-bb?an;-uW6GD~*b^-G|HsXO zr5+PEJ>QqvR-nCf&Nf$;GRa?4K_t^oPh6BHyOTQTLUDIdu zW&FA6s;gIWbhW4yQ@%9N4{E%w60$OHQ|I>7W_&sAkre&d%7Epb888T94n41~m(A;b z^Jlm8yVQR#FW)<~`{BMd?~he$zRmq=y;0%Aw#Eb1DZB3eeEe?x%YV;ipL5%&_4nR> z-iy~ZuH7LHG2i=MO|5n8o!sNv``_2u+*$we$KBljUnWZU1mCuk{nvD0Ht_ImdKI;Vst0PZbCfB*mh diff --git a/docs/README-unnamed-chunk-15-1.png b/docs/README-unnamed-chunk-15-1.png index b99cab7ba9f6bf50a93a179613bf5cd2f208f812..249a7713b1eab02a2f57d38dabba956345b8bfbd 100644 GIT binary patch literal 7953 zcmeG>XIN9qwuAsmKOzEB^wG$Eme7V;)|>vzBRzVCkT-FN@oy=QhZv(~Ixvu3r~`_37ZiLijQ z02ddRu$k%cb6i|dGN@eJp`d4wwkZdG2F_TVKLLKZxDZ@i9$W|n7XmyJxjZUCo73W4 z9taOm5)q&?5tIQBfQkTR3%~(_D-qltiQrl30T3!dVJ9NE5u1atxR)Xh1 z;sCqSgR9a50R~hiA}SL-wjdrHAmAy!y8B zZeas@)joaa`C{|e-hOPjkS#a$FoSBKw7F=(-6+vWp2Z&<-f?+`0&yq&qU@4Fn68ReH~ZTl zbTey!KLx&gcib~?pgt>!?FKQPem^ryED?}^6MP$9YPrfSb<@qn;Dl>cHl$mml?M;; z+1Fy4R=UxG+XA-#omyqP=M9WIWVO-76i|m;t9sHEucHIoig$}AEl7zZp`#_>gC2i` zqm@^1+UM;2r!JN92oX;a_zWXNGdAS{l_6QAudmk)kC6wrB{91usmF*r8NoAN0x9dX zm&nj9oHD$nq01V9MGIRUqazvxQ*OB8!~+HCPpx3++S73<|L`fGb!t9;K8{N0%ZVOjcl<%^KUg2vFa zpyWq2kyrO`)B^0WLuO|C(N7)uVRM#o`09hIs4J%VQGBhBO7O$4?^+3TUws}qLMIjJ zU4R52WeMfn!HB(t*%E}>oSkw5!V4M^#-1{p>>&n;yp)#T1Mth^(9R;v^w6oZsAO$Lv?P2Q_O9?kk^G&CF;kV4-~GKG&6RdJ#79LdyA9d9rw;SB8ziy| zFXU!?seaXZ84C#zy&U2F$jjWZ8$})zt)|+$Kf&8c_RUHX$b}6`>kj!AsR|et+XE8N z{CI}BykCu6ctN#Y7<(szz|-zz-h=+#AwRvGFWPReZ=u?Eb-3fsZ@q^Vxud;r4%+9q zA>r+@fQ9`rl5nPRqC>uH!*$1cD;-vAix7N2;4>PXKht`FtSKtJ6Cm7k$QO!r8%dk8 zM>HB7Tx>l4&>{aG<$Hp*Q9=U6S-%c%7XfDRw=TqIx%@bRC|G*z0)*4b_dYo{o##hc zSr>@_vH2(9ab2|R}?b)9EkIsL-ySvKXQ-AFhcJpuh8jzRM8EFdjqqv zO|8g255+I_#B`86aJL(m)r0W0092>=P!t2z#?Tc4CA*6|`eb?>56206XGbQ56g#uZ zzw5HNqYr%6WccMNX&IhRrk3vzL_8>n^dr`1<*Vmq>X*v!*Cf7SlbBApi&iv0+C4jcETgvFdQnikW98zu~Y(~9*(dEiy|31ZWy$INb{4&Gk2 zgBrW?BcTZm=`I3oo~~D1-WGPrAUDjsn%DnxRmixP|MKdvE@qVucAUurhVjN z4UYV5oZrc8#nPu?M>H6dwFfOpKzRenf0+L3qe?*>AIimtRXfYzj^?xH$F4^L_xDO# zXB9Wuy?l!~mfSEc1arNz=PWdDph+{xx)F#G1h|~5w7)lgm?ltWjl`nJTj6tyjT3q@ zrB4<4Wt42*)(=g|K#WZuXfZq+u#3mYU8#RIabuHn+Gh%HM@Yg06|Y0bPE}FlU(9W> zaRr#UpcLZMNoX@Q%;}OO1IsZ?AYd{e)5k2%eCQnJFutc?3mf>E4Ue_oCR&c!2?E8{ zNGt$6>^@)Y$a<_i{E1lXEzIHn0@Q@%Hj>vCqYx5H=?vOp_I2d3xrx?s%CK{(o#&S6 zcWlda8kS$m5EZ&+F2V3AS`ztP0xIgGDM%lgm4gp(om+3b?igWt+ z^@83n#CdwA+y{u0=@;_o zBK1`{UhWkCjQJPxihlG0Y}Y~c-_{@S>faE_?&i-9YoQn=Sl1`Pc6kGnd(Fo;qCD|q zT}VjoR1f)V!`Bsfe{6+=0b!SP4bY9ovb;ES1S_m~qCRx#8Sc#ZvxmT603{*>j1^^9 ztX-34;Jh|#_gp%ZJa36!OA3_VEdS2>XfJI0rl9_d;3{Kbyi?QivecGVEl=JK2%l0^ zV@Ce#vOn)jdy!$KOZ1v8W!q1i*~T5M`ojLh-^XkzX;VUCmak;r7dv4=sVBd{c^+Tg zi=-&W<~78#KGRvY;j1&|aU5-Y;qXCpt*Kk|Io;>o&G>zeAKD&a(8RTqjn$$Jj#ro=>rhE&5}LjBM5zU45okHd)j z(R%+ShoajiR8o;9hw6F3RnDGG&JQ zdw};c*h`k(b>qm%rx81kx zB1yzYQT$aj0y+9>3)nZyNPBI;@d;gTjY#h_XLy9-KoT~sPjl-{r|xr2DW1j`FDbj) zW&rviYI0Kk_Ha<|+x}uZ6}7SQab)WK;XD`JR7ihv@LmN{VWqMS1TCqmELdn>n<)A> zQPOz*f$C_?g^H25F84Q`KYi<@k>Sf{86ZTaSD#rtvL&nP#01!BUH-4`?%82p(4r^1 z;DNannkCal#_RU$Li!k|D_5Mp6?Qev>d>ynU^3+10Jr>rzAb#EjJiJTZfBXXs|25Y z|DE9K&mdV-RsfzP?R;hZ7WBSw{*eEO zLFbP4-RHtl#(D$OkuA4bl9&uKLF0q{wjUB2)@yqI*zn|z_DDx<{}-OGO0*>AyCofF zL2SOVpLR`qB^vb~Tl-(w8u741)%5BPQ^954OQ`Bi!bN*Q!8T5$WPCXstEe+Z_GzFfd3uab$CkeGCB-6;eur}FD| zsqf`{vyNY`Cn;t{>PZGsMoq~XcLC|kNX+=Njxr+nWQ&GqiP2)s)Cu_rz-(CG#SS)$8z$?P;?6 zoox9o5R&RmW7H*MN$yxkOe#REF1hz7b4Bk2tSDV0Unu8x-ZNF#a@O11wJ(Zv8Zi32 z*AObIR;9XF*X_$*QT!CK_?gE7affImSv!KymwOu&%tMr6)G4qN@OIF@MqWpO=>j~j zimS9)T}fTN&PS$Qp>i9D#TxfRFX~=rRylGj7#4;o5%9bgir!&bpK*5&^iq;|EebQ$xgvQTFjWSa zLOS`Zs)l}q`Vt;pT>_|+5|5KY9ITo|sP3541wDFRi?zHKp1o^6FM9NLz>jXnM&0MN zQ01w6`+@s_B)9WLAR0kPF}w`l5{0*svLlZgk(KYMxccv$5-@qV11Ram~#Uz@E{snXeA1vhnE!S)=);E!o0coH4olLxA~`O{0P=pYZ@G0q+LM zSK9Ovd+a@!ax{UrUBmmQ%uJcPvE81u^E=a+BfOB#A^f$S%WKu04%y=ltPrApsl*Xn z={#;ejN+I1qk)C-Kw?TA0CzBhQmrG{LjSON1J$ZR#nMV{**lulsfOCII~87UydO;` z@ut6Dv7I>5Y(xYz%srz_w8S~*>xIrgQz+K9r~9we3Sm2X`zwU8lmk~G z+XAyJ54O*@9NAXqI(o9WN6x^G?DbAB1KxH}P-iKjZy?N&^B=Z3BuL@amcIVRyJvzG z;((^`rW?J3>XzZ=hfqFmyn{k4eHkN5!9R&{ljd54wMd1;b)1^pW=e{Wf{q&d`qB2I`@&0`(F*9T@iRz%mc+UqVl8!=G2l1xklERJgr{J3tN;sZ9z8Q!C+;4y(zgC zf8YQuXZy+~SfF5-$((^m~Rh3DdS;&a~a0Y2(K4b{&vLRsFi2CUr zk7M6tu;1`<59X*hW%Oxv;9YEVjR^J?5*<=}bXsLdEi?4MW-J&2|b=4XiKYM%06=;|M(`ICXlS;k3apP=N^AoLPta4zfo=8>yUdIp< z_@F~vnmI{fWww1*BSXq27GvuZw^4oS=T z<_%db?HZcL*y`w9PxZ%$ui3+trV3m@`nNZuojTQbtmCb4G9yu$kG;qy;OrAqK2pzP zlZ-l=95+uL^;c_}ZW)>649j7C!-2_cRp~Mc9ntK=?`<|Z17kZ$JlO0!cWK6`N6evF zde_(Rqa-s6-PBJnJicqOFUGX)pTb>i#l((B)Gr`CGpz>GGp|Nl)v0#kEs4!+zQ^nX zTBOEvR>{R(!!Hw3IrGfuyPW3k@QaPHqKi0>>o(mz%k*wPlzc*?6A#-|l8nfey+_>f zvHjX+qsWY8+W7rZou*V7({Y_9|B|^FsvlVwHCLk#r640J+Im_S^fKFW{1ns!(%2<> zJwtmsp&gjm@=NYIO?QbMNleF|$nqAfpxI4hM)jKsg`q7bQvIv?B+*vt>24L7((?L6 z!HOB~h^9}o(Vi|HsLj5W&Xh6n;ucNXM{nv-^?aj)X7bNrtgD1Je0iGNYV*KQT7TTb zVex#s*?S&UbHh_zx~Kgfm(fEkyKs(=r{0wMR?J!L#FhA%r93vCVidRZ@y$^ zk6+)CREN6^tGYYm+1CtQ)-WZCHR+I#_5flJD@f9`23zI$C_b&-FSc)NH%9eF%fWbJ zo`_v1hP~-S-i?~`H_+TvZ9LS$mp&q1+@?-*jJJ|e_EZldhrT!YD(8Vy8r$q)=H;fC zWtnzad%Uhb3tjgWtA3N^Z zhHMLtZLD876y6dP+czqWk?&J}C7a=rFfxBmxZ2rkOpT_Jx8kKSWx4b{Fgf{4OK7Zm znXr`YzyAwX-ymZs>6wzeA*-anoK0`o*%?FNsp|#m5n$P%!d|?!35ATGCQtiR*+j3C zHokYKJ;M@kZ`Pym>IZ2-T1ufvOOk<7?8lre80`l?Hcn2WY5MLqtD}7fIL)v}=%;zs zH72YWWE^RA<=MvZx~1;Xi1dBoI<2*#SVtC?@NpWmF}%9;gT7yp+u21j2vz&FVChHO zAEzM=Qwibr24Wa4&>y2+?&MZ$}?gRg01(%+_2b|2Pu3%u`45Z7sG~Tpf)W zr>B~T!noy;?(ogSTl3FQOU5UDZRwtQJvt3RX6cJ#ga9>d9{ma@?=!5$=r~&|QQmYU zq_|jSUKtu5K4%3fhHzS1ZgEtdgnQq_F@a!!dtrAio==p$+9Zt~+K!DA&J`V13ZN|R z#?FA2lkP+Rh(%guVHXtAypI+X%(@&6JoEFO`9CJ|@y22dgx-)YGJ|pAtdE7rrxZi6 z{LqyIm?wBADg!0W*{b^gN~_SqsoyYS_gu>Ej3=6i%5N~^AT}}9&OS@Ons7S@G*b;S zt6y}i_`G0_BO|(Dp`E0`x*gL$p2BE#|JJXbtO&*Q;2$UVr15Y*ZJV7y9WOF+z4>3L C?5g$v literal 8047 zcmeHMcUV)|wkM%WbJPK(%9Wy0f})gUfRU(((nO^<=>kG1p+kZQs3=vc6cGfG7U`WR zA_k-xDUlWgqzi-^2;`jvXXehm@4N54`@VU9y_0iJzO(k)ZSD12zqQYKps%aRbwJ<% z0|Nt>*5wNZ3=E7|;9_871SG@6Jz?N)NMHNLMc|)-0m{JOzyO6ZK!JNKgF_YYOnago z4p0Z+jD-TiSl}FT092vCNiU!k0!J+HbBG1*RStkc6>w5xp$xGOzzO|P5(^wvz%RB6 zxDUk+QL7vnsvMv|g{oL+RjdQO#DP{wJBEN?)evx}R#B<7K|yV)sY9{Qp;(8ZD&UNz z6#~WdYiNi%M5O}FQHNrwLseAZnO2D=>#8{24or?6aoNO+fr0-i?E+~b!=5lOh(NV2 zoWJ3hzBm$Si#Kd~dFv=Rf|HXyys8|ocxUruis}NM<74_+Z#b?qh^pr;KBOfS`Qc$@ zZc*0iFH~_i=i3tfQLYwIkY9~kb$A$zGtgr4AhRe>`Leuwt&C+oEU>I`Fhv`ZS&|>vG8~{y&UOvyY|aAaxQtCI8Ews z=PJ1?ZSD5pS}!hf1b+^rE+f*I736rFR|e{|k-1iPNKwD!9xJR*2^?{8I$vBOu_9DG zqP1}o9MRFG>vUQA4x{E!-xXslW@h2Z+X@3E{o;ElK6&$svCad^yK%--SGEMQ`k6n< zBJ>vxgNJdsFsEnc>>T`E!rW1BP~Hl&astZR<3o8%8_b+{lrs*3uYSR1sjH!k1NMRw z!Z3N|eDW2sC80UVtE?K`-0h~&KoFb*Com|Qbe^ld&%O3gX3)?~jUEUtSRkM*#4A|Z zB^om5RyA}L+jHp^3vZ0V(IrKw?-k&wY&De6sHXG#B{}PUE3{dmxf-RG0O|#QG~VhhN-^8lX<7gfoee()nkRpABNqF&nyWH zR2-H!xmv~2wcCGcm5DRTv$?W8eH&f65|e~B5haY7PHEr44)_mHo{6@pk*^eu@9zC>~sGl>1PFP5-!8hX-Jv}sM(DOU41{GN6rbo_o2v8x{jkvWKcBZI4 zCQbN&thG}s;W8~GW4HyBLv|=+b%9Jg1Zc*BQM4+`W+E8L>>~L3nG{YS#RukRVG%?e zhu6nF<0%k0QQx=|LnX*5Ry}M^o*!LN7ybYmx!Q+)SwrNJ##Efxy9UdigUJXXv4o&<6xFE`a_3 zM0}BbsM)N$dq#+29ztI93ZttHVD~V7YayhJ$HqMg_qd)&kH67fe~iwS(h}UVH69Lx z0Bik)CAz9G5ok!NK=7sE+hz{_?YM?ePl;hyo=@Hz;0RvmK8^!=FEVPX2Mt;CEz`y6 zx@@5mcZl7hS?V&X)F#|{WBiXkdppRC*^aFCvKfu&>~K4-1)h#)EK_sU=InY+!M#qj zHVw`plw}kxGt?;<3rO zk2{6|f{&Z$^+W<)qX&DgHe>s6*)AgpIIS@KSPd zKTvn$B5Wwi8-vK6k)}`1o)LfbdKcKo#AL?u3%byOnX_@$m207{r>C(?jWNcYPXL34Bz+JF#|Z1@L+WB(p?Q6o@p-wEMuUAplqB! z+CQS%$sNusU-8vdoR$k2zh!w(g^>pX*Cz0hsGZV?O~aq$E8>cPN!WNEYH?NQC#t7) z(p>Hp%lrC-RdgYwwZ)NV@zPzjTxVJ-47p_&z%t+^k#m!!;I`@D)& zy%S!N{7otBk%U=0PrHp`6Lo!a#Aca7#*@w`L>c%9V(yv*&hnC47CBbV&H-$Hr9GT7 z(J@|>UQc!MLzO3i1}L93?cBmXbC9?2`FdP>b}1ieb6CK$11u!<7iEDKGLuqzO4Po6 zXt4XuN29ZO3V2BqdedrdBmPDqw>@Lrfw203my}TaHD0pt7VrRgl7Z}`X8pyLnSlZ; zAZ&I(nuFcsX$%LGv2xEG(**lWYoYyxw7TERi?^{G6> z3rN>BEa}9%cATtXS#s~`;`#k(!OG!ZOH+6r6;7FnbQfFzXVRj`T|B!OCYOW!PgLtB zrXHYzI<~g(tmp-udtJrGM3Y)Me{k*2b2^%GO+WBNw!SZ1jRXSqn|VpPrn8vn&zo{V z--VZWH=ojNzE8Q4@M%W)Ks<9pic2};bAaZJ0Tj)uF^+8><==LYKFgHkDyD8c$YZ{6 zHjRe(lIV9|_Jiy`zp8pD9%AZm)d=~)sQjig)?&@?&n+@>7PIk#=T(U|*ZxbST7HPJ zRD0=GRyzw$C2j9ac z8+y2soBZE^wI9C^))kGtZwqI~+8&q4NCV`zt`P#D)uihm(aOg`#^F2jk|z(!SM+LO zJ4*|W1KNFn_HSvRj$9E<)T2i8rTNg_uPpn41m>yG7VRND9u>i{dA+e>w-sIym^W7xraa=V-U_bnZQxL|k4N%n9 zt$O|jw==W=0mPPT9!qbhM_1fDn)k^BChF%A9;ocCChsk4f{m(GgdH=*_C%bMXG;36 zupA78TK6gqxHXQK^jl6l_7)AX+FQE&R752E2SJu#a>ukab@HIfyaD*vJxC^Pjjl^` zL!YeX@>nm6Tgb?@>A1Z*^Ry9d`zTEtiK#K4Nb>qDfYHSS(CR@_b z|4XYqCGSg#Idj9!nxcgPpw&-uayicjw@T`gy8H&%ca09pvrAZ*bH8%V7tHRsn{^gQ z6llW%WS9N~(dg68XA(U&fz8N=~*O5TmP&3A9gvCsqC`Ssi zhu}!O#xekKud-H=5|TSju;S(Yj>n5dZatNP z1ec(-E@6!TAycfUhsxerUrFj#x8$bLtRk@wetUJ2TZ)V0ldRlMU*y==mD?>ULJuUY z`z^{^?cFyv5c2iKFVV1+voe<}YlM!pzOUC4Ei`+it9Seef;Az6xqvtMng@y$x9JpU zh}qYxsCne82f>~mEsoAVf{+4~9`GjDnit-+Vyj8eYPq`x>Mn^y7BZH=4qOKewII7T zjMJ1FvVT(zC3d}M=6hjkzVQtbiY&Xyxfg?m@$;EKarP)k+ue{E&ndR?^q-k3=Uc7> z6eY~exg%Cvla`({55K7)KKG95SQ+Z9wP|x-ZIUz`LK}@1QIJ$tIeIn*ozQK9UDL$kre&su-&fQ7)H?Me&)uu;?x0QsSPv@pcyn=(j?q8inqJ`Sq4L|NRgR^U#+pCGNB=~nbfJ@n3c=KTT%=@^ ztI6HT9B~aut-Km3R!u-Gv2rm-#GM0-Q!PBuJb%_twnpbuFiX>t0wT?$Pt^o#?j2RF z0qJ%4WD#AcaMh|nFo5(}2YS!0dHnYrIcUTK>4@V%!u3i9#&d0F<}Md?QexhjKJUz} z7c%lqYS33&n+y=vg+_2khmU&8co%x~Q7hayi~v`rh80opKKm6BiUuSVUqM7VwIKf4rmFr^S#n~;3wUK+^|^*$c;j&QVL#z zDcuPQ(vw=4!`+fz`U=HlOFs(jDweXweOzTa_14Y!(E)_1mZ;^-6YA%r*r% zws^B1fBvD?H=ZybauPIf2tksH%vg)(d(|Q}uxkJ=N!;cWq7t?iSM4^yU5>P#xwFhS zud>$|&RJmyAmizBK7<-=F37FPI;$(#N~UxQ-4N&p+foq!L?h}8CiAOUTviYau+B}( z;4W~91aVTYDFdipKGD^W_C^8$u`ubBw!Ex*n6 zAzr_9L-ZKbO~LenS%Zq35h#el{HYOAM*irShSfYIh!LXfA~F6oBfm(=2DJYBYriCl zU4&?pj_(;peaJkXJv4xpnQVH|V*imgoYKjX2h5+Z4s~zoRIWB0@GHEG@YQ%``6cAi z#n&qq$g2@6ZV%@*7}w)Byl81z+>PJ2P>YVVE!5#445GRB#}-N*4@>U^>w3^z_HEGsRtuJ5Xc!aIDle zBIT}FiQEciG<_FO+JZ_+1Y1<0D!lf^>zL>su@soXAq`=iL)7`3-Wxiw4P9H?uZN&G`?HUhck8M5m4kmn6yuea-{&4GxT zI=hxHqDjpYzeSJ7K8yqH8!w6*4KaLsnL+&lN_>`}Py4de%!W{3P%@{ew5+vxQ)BR^`IE7fV5^E5NuJmVZfdVhi= zj>H``Mn~Mi>mPjXS7}t+&|8n^nzz2Cj(Cj^NlV@9{6-&X87+0!_JAJJFAW27_v^sR zF`2k|7~?b0zx6Vnz!J1}HcQ$?xH%YcOP<5F5H*<7GER8$tC!H%-i}+>hDTHyhYh0P z6#3NodI#Hfo1Z$*5yNIHR}KwNU>b4EO451`F_7ruy<0Kw>m3J3ogVjh%GDv{-#rOS<#fK~y3P}LL+)BW$Mp#KFT3*@dn&6! zUw6sb4$AL3NSs@CF5B+bWOti+m`o*d?|`uOwlTliZc9iTIfxe#4gV!2(0BQ7-C~)Y z$kI0{o+cMoIky&KJU1NM-*^l5z8#zs-EbFfQdbG0P-J zY+!z=PpzrHXN#?h(xNSGEIZdnxJYM-oRKxkrJRv|8~y$pnbP~pIknAQohUZ7b7_>E zdikpD$i|b7n_n01I7h&X`mayyvi7;KRrc#k#R?@a~YX2 zs;jz$mfCT8@BZ)+^(e)*t}p=O^+QciAnUvE#}#BoLui*85ifh8@0fy+&AIZgOL&xR zW)RN%ellB4>Z`F?aLUAdol=I;D5zUaLAQDxFz%dk^aAy9t@r?Xbx;HGtzk^4`i}P4 zuPB`V@qSlEjh53RALrH6h7{JsyXmH?bZiH(^pkFdE#?kd+$eK zOy=uYg#yeuI&Zrm!*)$;DYBjahh}s36X(2=suT@DYJb|Z=-A|l2 z2}YYIo$)O?Y-q_^r|EgsW>Xz5^NN*x?+EAxIW}fU^rSq3Xq;+DRmtC&eyu@ zqHcvH99|ZHop{LS$n(v{O`G<(eQNC2?qYAANM9?RCh~FxL{`)wmUK*%P@vX2#p>Xcr!41Hb!#R@Tif? zbO(q}aH;7LRb>aeAnU!s)X>zFWR*ITK;Y3EQ_vI1Wd0?%-D&$Kz5g#e?U^~{;(ArD z;vuXQyA6HpX*R#$Uum60YIrXvbfs3wAJJzx6Nk1v zv<$5W{0w;YiE}1*OctWyfCNTRQ_Wbuea-bFcW;XNp~8Tz&2B3-S3WDQt-?0XxWhD) zceEbfhKO+NPC2=dh+WiOXKY_~qN7>khdA`Wa4!!JuUvtGs^IL%Y_}Eg?Pj9eYV7{H nJ?aFp;ji20|2ID3?xMa-#VcFIIsHugaISSx_dEFjWBX$nz9dI_P#hzin+f=Y>ih=LR; zp+giA6-h+8p#%`3gepR433CE?ulK!wX3bk~X4XGz-NQ-3{>t88-{qXU=M44vc(?O{ zKp;MYlPAuDK%7M21A=k_kX|ZF3izR&J7cN~{DMI0AdoXiT^*zje8+*D2|%0OVqebc z&cGd~4glkTo8r7i>Z}fOUdK6u*m(982mCq50pA2?fQ&$JCIH{7aq6HrXW&+6BLVc< zCoYZ>M~I6f5aI}wI0_+-vPy6U5uDWlPC}eIAR<#-QKK+#PeC#}3eplvDa4~Kzo z=sgff7G!Wj#}u79krsG~qL+l)QNL}UA>|VJZO`Sd&MGyH!?taQCk@8W>R(4yU3w3j zVNlU*at~yW8^X4SnV4i(_Zimn}}`|5Xg`R#~~R?1jXpsgPLRy z+_)sTjd&2cg}M#G?Xd;S3FrO?{r|aH95+j9$PoXC=i1A=gCi?0DWiX8ctUxlGnM+*8&ou#PZR(qc_}Ygyt~X(D z0YUin0u%#jCnE5ZM~{m5BxLCaA1`z;-zT9Q*;KA(MCyMg5Je1ltXOpUi5(0rSXQ0P zonj9maY76qH9H1?=$NOg23y%<sCabDuFn_zXw2sff5u7fdz< zEll4#V;|#_%>kz2&QO;!wpJ9LR5dj+9>;%x07Tl$R4G zG~1AgoH+SR^iC*_Z`;^ZM&&W;8RJqAZvUqlF`PFSE{6k*;sqOVLP)}pbZ%(s9w>rO z5N;xf-Y)C$RCfA^B3FSjq;4-)-J)3EOBh@ZKqvV>3G!TSqO2}HK?h_9*(pI{pkSlW zY~n7Jq}VTALUJA_0Cpwi5VUn~pveT}+Q5UYIQg5zHhWJ8&(n*S5Mm(qMxW!GqX z5js>Q9qLtkFMWmxR1yYABFYH04JE<=poiwLta*~`fhaC^CkI5?>envg#2KBYR^MD*N7XU`wCHwwOteyq~`t;JiUUJ`XUfR zaRD_yycok~xt7l99y|kmk;x0ri3*jd9zc9|^C(0m^!aY^zl!d2PxtZdW0LxRjSql& zz<4m=z_nfkx_P8|0qekdrC^XdFP|-XcPvU)pm#**+l!b9qH`_N?)s7*S~3BnHBv8p z^CsOEWt+4>FXAto8Z%Hc$?h`gScdKPHI=4KTh-%-djf*Lyd7|!(|Iw{Xp*o#eon-X z=l7DG+i2~K3B;3UZIOd;Y9UZSwOdW)QnxT=0FOvKI!7Q4d9^5lYbdl{~_WU55KIM z&M{lkw%-=n`Os1fijZ8lx|J{k>$0Qr=n*je*-fH}E*XD%@Osggb>|w1=we5U>0*C2 zTO!3Tm%nmpk}ry2D}RJpnb3E)SY$)bLGOPZp`(%uNo&2{X| zm@b4OwzAV@HG(Mr7l5GSx1!eqhNSO>x`sN2o{j)aCR@Y@RZ@=7G^^9Gmk_kPARL$* z6iQT;EvitQH(lh$UDvkvG#-EcgAYu(K^)5LHW`tC5IAs+x-WA+s6kyh!I6=UM)%zX zySX8vVvXs02ma~s9~=HVgw3<1*}s?VJ;#7bKqDtx=!WXcTm~}o{4F3sTkx?Z#rsJg zTHGsP1+3-s!a(xr0=ZbaQkgEouywCr`B@4Nl)tNQe_swroahSUl z(+HY~gGsc0{JoA-8IjW19DF^KNdLbGL}%FX`m5mf11cPAA%IUsgNq7~Wg? z2zZ|uAb~ooeO{E(U{P!oY8b0;pS{5lD5vkqs+yC6XmLTz_aCDk_u;BzR52$R%|@JB zr-!^=^R0_Y^*RO=l`<$!!1E)%5y`#f&B(3&>wwna&3)a&_tUmi?ioE1vwK4|x#`El z78eY=xp8KZf(wE+*nUkcYlYI&(?#wIGJD6BWcTr$&(R6r9R78!39Jibzg~=zv|iBB zK|^jZ{m8-%IS48I4$@slYb954QtN`Za$N#$|1|%`2*m;jsIr1UGp1lTYjM zyKG`_l^0S1o7*=q=6GlSasO}y&p%O_nccN84l|rl_*(1~L^v|^AOzOR%<>OB_Q+zN zmNd>RLa<_QM4tqVN{-}j^bYM3SX=fAllg3sz@k^ODu3J^$&}pr8)f$lt!=o%LP_Ob5c$Ot@N2t|(A=Y0@>|@QcQ3t8jsvziCl` zbWbj^al4T$GyVaotFHOGRWya@i?btm{i;QUK^c8<@wF^o7vAGw!#RW+8d&iGJBQ;l zz0^5QFzSsKw&L;XwseV_5MW>NZK)Oo=K-R8(&h@Q z!b)aC3#!YLjQ>n?5nETYZ-rMFGb=Pyf9-x}kPZhl_BD=-zClCONVV(++7-U<6OImmjL&oezs z?h|(?pFQ>)?4n^fv1xk4{_*>#=eFP&)D2DKmx;53SXU49?|m;fwb-_URgi*&RoQJg z-(ShnJ$rPz&mDzYrBS>RCe(<~7pA|&k3qVti3x=WQTru-Q>p-oty#`ZrB>^(;5)$k zL^dLg*m3e#I6%Mx|2h3inN~pjkj$TWtgUo7zh|WXU&-)zkX*CaI`!j(lHu3xu#HF~uuM_Lsuhp=fkL$R z*vrB8buG1F-Ag6j6R(53{J)0pH2sDEybV;L$VqO0jQ>|{`$WWTIf&MqxA?whMvlMc z0dUIKJ;Fe}j)5X-HN0et4APImz%pQ*#Q*lcnLDO|B5M=WY-sk`sd2xkK8NM-$Iq7aSb$JG4X71*3+EOXjQqm zU+r}V)-GZQ1`m=3qi!W3fjOq1K9P)17e%@6oPkPe0oojE#s_%Tmkm@`C`zCFMVf$A zGpds4rT3)SL4MJlr4|%a5A;ZEjN1*C@HKcH0L+M(Wg@V@RTbyOm4)%PdE%&wrN+rS zQ-Ki87$uw0ys1OC6*pwt*yp|<1?1e8x;dCs`RS?}9E>_KofjXW?BK{$l>}cC5P3Qb zhFMdkurN59EqE0&vOZ{edZn5X#tDP_u|b`8CCMaun<`5+udIfl2UPtvW85>!u2Bjc zp6ntgiFzM($YqQ7`@T2cjm=y@oNZZ5Og;5*t(8k`IlUb?F5;5mmiM^DcB>Pv+(H&u zDHOBWbHOUg1RHyZp(@*0fiep-OBxnj?ioTYKunzb`(Bq)74*vs>w8^aN3*hBbpcgN zM(zl37@5uS6Ey=Yd^^F&J(G6}6i!f~>_NV;lvvj)X@Rq}uHwzSF!(V*ObE93gx6Sio20O#<-f-m zfyFt|J}z8`Lk#^CA?7gn&li>C;@tm0pU+UU5oqvjk#VG8%BG! zUslIp*OR*_fWZAQre9{g7`dkB>pab3G|77`{KNK-N&XfkSNZ4vcKX4O4+8)S zu5AJq5Zwr=M*hdUJuJHVrrHBSA>5mD@tAS3gFAeLHMBq5*dM^*F?!!7IFX+Qc({dt z6-a^>FW+upKm4ra>&9%KZq6$J%(ER2Eos|Pj<`+0495)|TjZr}UL)KY{KwV@Yx}+b z;*%nno&AmPhx%o?HaETDJoBiP`|1_=ZW;U6QDP@`URPXM5gyWyi#NT6-vTj|zZvyJ zZsoG9$9CA|og#w@oVd1dS^4Ksq&PHH4=eTSzX4APX`bb;mPa+WsY$1?v&ze%_9Tx|f(F(%t0U}{F( zv3+=BL+46^J$C-aXKCN_c90KP#TIXF7OO*7FMV-TSBI(RwTfrYR{1_1M!(%eAJn5Z zdbPEtlDzsfkj;5!7Np4K=5q6qDY`>_EYFDxT8uD***?@-@tS z=3Lf3;1NvO#G}Hx-h0TKY1sz5TX_t*Z216a*?7F%chXFS1}!gNr?;G*ODVtJ z6~uI%?V64A-^nTW!S2CCXysmUi;@wJr%09sb1qX_ViI^iK!N=(!1KI=I~y)*s`_=P z^!`U5?4b|pQg=mvf5C5CI{sxmoe-ia(3dT!cQ0kAMCO!0Kfl|Rt3?ki%LizZ5~c>(Q)W(p&`thGi2jX)pYS2Dt;?B*990TCPaT z1tQ%3cp|==Zd^Cy-+cVJ%-Hcf)wb4%b5_xliqkHG-fgeg z^`K*$Qqd^J^;x0Rp+_;!w?4S|=o_oYUpl!iU4KaJ!`e~S_xB35@8wn9gXbs2@Q*xu zLIO3M*W`P-_%Ir=%T#*JmZgT-l5v1BFUxt*htm^{G{p3Mj9kyZzY?VS8Id@AuOrU+ zK|#|}tNnTgO6k%axdYzrp~v4P=5wBrS6Sr4i9Ta_ch?+44d-TE3l?3_UGdNVzsiFDRfY0(}rZu#IH@l=OmX|+6^sr+=+2Qh24L?5Un_kmjC`5JRs-h&?8@?H%I(^u*bHe;pA zq#xjZdkAx3tu1n%YNTC;tgF5iA2yRj(_);PM-E6 z_L%nu|I{gcW9B(OIzNvwujvhTo0+QrseJWyPR_{o@c-7w7+}R%9 z3qLH2-@tH4QVgLqjcCzqQFx@nlG+T$amg0#$tLc*UVq`8Fg(-2=rm;$LS=GWbkG9; zxeGwr;F4E^Wi92-;&v5YNNl8lak>DM4|jWNlaMgl6WC>sXfcjMl5P;69^op0R0B^$ zup^x#{V6{ruvl|9XOdlEx-D=zx55e|X-zl%NI@Zk>Ml-%7x3?rvMP>W`O%z*862BQ zAxjQ@h4uPZ&IqfE_+UYM@NW!qe$bQK!p`d;q`qhxGZpa2y;HeYN} zG6y4!R5%UJE=fek6Tm8+&F1x+w_htB@|ycOW|RNq?3^`qo_<9IGyCL>wml5$`hjHg zE&B|1sb-;atOprfcX8#qc|9^MSt89#NNwLEFgfp0Y#wJfz5J7Yvg^)dn-S~!k>MJ@ z(ppz5pZeZzWi3)}MzYFe|1zZ!E1b{x2oTPm>dSuHtWtihICJIgXW(cxxx?^_X7@?W$SG{yb&txPM+G6`M-m^=nOI3MPd4FLCe(5c^=Bd zCD$~-K$sS5@}veLj3_7AeQ)%xBr31(qf4OGsh4i6PKVtEeyR#M`fAlZr_`VT+YgUk zGV8J;um*;vRu`|gYWZ|Z=f+%-j4TL2ji<(YmLYiOmi-=@e|`#_aWHy1GgAfyEkS@! zMHU?T9@F*Q(lUK%$rDLcJ3JV}Dc02VT@vr>_q)RKmmvLxU+)C`LCRtT zb{5J{lh*XKm4b+}LsG!e>?d!K<~814>k54;=Dl9AHq|<{&<;B;R`GIg^8Q`rO2}F3 z7=~jJu|?NHwg@Z-$Dcvz3sB893zj9I$na6bnrIz`%Q0TR*7ADtPDRX18MIc~+ z$HX)v7edp$d8ymB(Zx$Y780VrUc_lMyV>>)W!h0S^)qb1YFLSTx6>?oMA54>)AIa= zR;UsfclM~#^xTWGfQ#6Va?6qn?*KkcaCvfHxc4r2co}d|HZO{nAHZ<-1>8T2(;ySX zYA@CT&+P$2NQ$BZR;~nmm>Hmx!z>goPNd!~L*&kSI+3jNvQDceFrd9w1lv+A-5*YP zBJlLEj7vFjvbIoVC#Ev79ab^xifZk+rftswqyr46h7!f+`$y!d@TYISh>xj|ocsnz z5)4RkQHHOMQCQMhJ;Vh2lW$K>a=IJ7U^)9UtHA!98MWWW{#ds_I5CfGKNV7&>$UDX0BCeYL l$?B&l?)Ja&OM@#|4}a#B6Ajmd*>8^;=o+3VIA(wIe*hAuaa#ZY literal 10911 zcmeHt2UJtrwss5wf+&O{c%_31D$S$Pdl9g}0j0Nq2e2T7UIRv|AWcN{AVjcm=%5fv zfJ86{6{Uz2Nq~S9r9*&FQvMylbI!g0xa0rd81KDt?-+Np6SC)AYnE@$HP>2u?>j%6 zo9y3vXfFr^+HYoRWCa4T;eihb$_7wKWQG{9=!}A2O_^cxx`14E$zA>IaGYrNP1AH^nH9+Z} zz^TDv0_^QidO9&3lb(*jq+^KbL`*u7iSYzsJT-uxm~;(Hy5}wv0~|yQ5jdF`CK1DA z)<;IRJ$gh;*C3{Q5;4FD^xkg4Y6TojB9jToViMC?&4Fv6-!@G#10e)B%7rks4*-D< z=CD3{%z|R>gFuoXGb8=Wp^xWsZ(bq(j&YH4RJlS>@2eroU375lFTCdK+*8(haaNLrmdGKm&1oS$1Bjw(fokDtJCQCvN1VpF&_!!#Fp{cAem5Fy3n_ zLC7Zyqq(43=U{5W(j(>oVGk-mYI!bp#ZvU|NpYh8NLv56&sgG-Fvw*wxihY-R={Wd zyaeHvYouZ&yZ8~>*k8n`acln*1$DWI`;rJngz6_beX^4t+445NUb6lWbMtHGsxU=K zajSd9(lqLZ-Ci0o*d6!&LjCc}d9UQBItzV;oR@~)yVY`HWe}!OTXpWo7Td}sEB8{` z3%b@Dbx%f@Pxo}nt`be4kHw*)Y4%HV8*FiDA0I4x*$5UcFxHVvHlH#lRAFk`Fg5X> z%>GL$7b{nu-&eJ`5yQu#9qP~IboDwUVIw`>N%Iey7yE45$O)lA zC8IUnkkzRyTGel)CaVF+2@se@xG}T%T<8hCz7V`Vw!> z*5AQ@kqXmAv=8_z9YPu2#>*IYpmD=_HFtAyK^kxqxD+h(8A8=cTyoJ=pRA<^-%m2% z&olhCMq?{dI|Fx*4MhQ=4lJ&QWWR>)GnB04@J4Zd<+I~}m>-11OTpBDx)QHPj@5@B z1fv8vP;Ssa7&hn?)f4Q^bo9&ivw&c7Y*C&Sr9X!m?9%vsS3P<~X&2KY23?@&9pM!L;LCha+JpWy0=mboS7f?5uJ!bLgn%=g+dTkT< z054ef?t6iu8$0aNd7Mq)ZON12EoSlqX;1JbLz{b1fNL@zgG+B~vVU8vqtgwx=lY5TNgwbRrn| zHNXv)-Nzf8z=A5YQga^)&(#_;v<;K^ERP{e)pm1%cus694Ek7EYe0+z=;-TtWyn7@ za3@yejN2^CWZ3`T9{o>sFrsfPQwUG*B}+uprw^s!e*)4dgEBSg(LJ^lQ5y<*um@$5 z^iUUl)WhQO{hSehna0O~n|dk`AuvF{uiKwFw$i)V-f5AX*{`?+m1hGyJ{J9g5Yyqe zK>PIFFon+!3dF~NLCT}5*7avbNQjQ{g(i-W^vHhFM-i_59~Zn+OvqBAamyzR$WCzX z#~RRm5$T#7QEyLH?gf65Iyz=+eG!t91y|YHMHOk;vrRR;uMZPj>JB5VOwt@}RRRs*2jplu1G0bl|P*)1vE-1BVm$Bx>kVuvpA>wfU*nAkugQX6kdXHz@xLbG) z$wd5&RyqyNwrh z{z3D$Y01($(Z*tszCEbiWcP~WM3#B8PJ$8EUpP=fdhn@&fs*$sEZa4O2`<(ceYhBG zpk8jR_~m|V}ReuIXJ86_x&s%ko)~SCICps6D zWC>QIQw_kl`%I29hCb$Yq;Jq(4$>G)nCkVo>sny$Gj*hW{T)&VJmVktc=!q0Ds zK0Z!NJ}@E(ja1%IX15jK?5z1^@=nwr_8YXLRnx?gmjtO2WokXmnWH?YiCtSePgd%> zMm5_#j9&U$Q9dnGd3i_rO5WQgsszvQ^l+p10#Z|l7xW_tK+(2HHvE?9PjU2#&CWCr zIVts{(`leW-QQ}WDB-I8h6P;mz?wIm4e1-#{FVog{OR42SJ*nX@sV3f+n%NB5aKB9 zW#~8lKcX(VvqGK;783@a--Uk=jbk{`C?+|&xx%l)HV5ZoRetbpr|Wv7B6c#*;63`< zQu-CL&<+p-t9DvEO5cN6s}qK0!0%dP!@$R%)d$Dm<4K3TwaRlNG^^MfbO257n!>?B`hn9wHUmnG%W z4dR`Ck=I}A(`9z0eA}XUR$Xu3zc^;agAU2~{jYtF^brVSUo(7|ge5sl3*9rmAg#0H zBFac)dwl87`W$W}a;egc;dfXw&TC&;BX|ny7m=n;j%V-`9c!l;osyV4M%kFD^Zdci83oz2{zSKR+52fLF!b;_91+K3uf|W|#E7mT5 zcRd&One$ga&3uCe>BX1F*RER}=Ql==>x3Bl(IGq(FnGIMFKZBy$lG5=_f6JXZ3xCA&gzo%}O_x*} zb?-E|o%+KXKFU6jK5oRh zSZHI~kE>3h(Eb=V<6Sr3=UpKeF|h`WA3#xBKjztjm4;z`xO#~lYrVrH))~~yOy&}@ zK7Lgd50$?RIP1}hR{c0+G{PRL!nw1m__8upQ9E+~2DQ0AK|ox5cV?H~irHBJzy`8+ zZPgS>|M6bSbrEokz7cD)d>{$VH=a6Ng0ju5P@F zR37@0hNCUFRE^b;{wr_uBiBC%pmoqozJ4@UQ~eFTxTiu z5s6dF_3N!rh|n7MN9p_jxlfX}Rl!$MH&HfqO{09CY@)(EikIJbvjWZ)5)N zMv$ED%tkATXF1W{gnyaP>>~jFImk-qlIKEoOjBKw$K4=j8Vq2mxI>`;{Cwz!zGa=zraa+B{?H*jSFIwKBg?G4J$@)CL(#lj_Co1UydnI%aoqPh zy;z~pR#!-BMgnkuXugvsF`}!nH99n}2QioAMFr^pWgq8|*BqQ|LY4~yGNL8h$|Wns zk4g06PVQ;Q?eb}rs$3~(33QVDg+1>8A*5ZPOI1Z_95#EengR?}fT+@MG1z$Ds`C7& z4nwjN>_8pzeCz5ssx{{|R8*Fw$&S1#W_oAH{A%qe%r4y0O+yh@0Pa-kcZ5UC)maTT z53awP)_ir_`CB06tU~2L+2}FJqX^5#w8i;wf2)z$Ky7<$qmnr4>3E6*ys$v|W|)=% zi zfLUM&6lV8)N5s8v^DxQEbC%O#P_4^{yN9Io;8_KqDS!ym0<7UJ{Bh%=DXGPuGmJtK zKq%`Zt4>35zVw}WV2@ZL_&hHGZ)nUvBnF1q96P7Yi%LlPkD`Bl=pLv4Og7z30HI^| z{#`E4k@Jr1<{8_hS>j89|GA6IXY9cdXwqEmK@4k@@H{NMkv{z?;VYF}e%*1OBWoCO ze&{)48uYz&2`YMo7sWRD{m`?9mC;a0>b(T~9!lVTE9Xsw7TDySA^BEV14oe}QjIn2 zr4V>6=a|dcI-+TK%Vn4vJg;28_`F2wcFhu1w&<7aea_n^gSh_}rF?4V{-f^y7tDLy zTYq;ea|h&D2kd@+$4?*I3kYASK0iYCZoQ^mvWOk|>wRrME1#(LltusLIN;vmPS6bE z=afyas{g~y&7EM274_ZHH)ww(#H-QPL}?$(KmVgw+oLnTqD&?_&Ae;kb!6L` zwJ)Bic@WHsVWy=wr#qM$G(yQHd` z;#;na?K%6go5SpC=mi~oqPmOZ9`tvJ#2Xv=%amwX4|cmMmIkVXL0`U3K5G*VkV ztZ6;kp){aRe>zaKxrOv z;}iL~yTc_94!&U|vT?XskV^_)otw%g@HBb$uDs)}i7jF-L`aU!&tGhME8CmR$=jlB z;3h~3e^g==XSpJ?Eua<Xcdq4%G>?d} zQa+Xi``_LSC&qBNT_7)$%bLzr$WnCzSC5C3+050uv|3!|pMODRP}|M6?k+9ymCy&n z3#TjfLyfQAbszLLdy$-7#xlz7>9-?)Z&W08%GpcUbEChuq(5-AXAONdO-$gxh2l=& z{QxW0y8vmi%blLoz;dHE-Y?Z!SNd@_t`cfcG&IHXbg;FO;}yZCL^hLdvu_Gdsl~<| zRk~wO^auDu^>1V;TEGKWuBTdPMi*$O$%_iSzqs_y;yIUZWJI*@o7J9)<66k*aY-j2 zs(!uK2NE=%elr$5nRN6|`~hF>$pl4fi;787{9Sy6fQ{_ae&R285~IzFLpab9>d=0` zit5xWM+FD(UVZ7`@dHIpfxj zzM29pvy8MgPF!U$5@pU-h1t|Q#K+^eWWsfWS}XCkCEVqG;aSnywo^-oA;ah{GQE#H zpJ?ChZ6`DW`~EAP@6b~_`4Wu>nKN`cpUGYKp7LJ(iUPdjcK>pGucEbWMaP*N3Aup? z3%h{z<8vm}o$1=@*J0&APdV2GBH_ztYVQ~86d=H<2MG_TL4}ST4rcq@B@3vRRA={+ zoYtF~gR7gOBgogH(_+sg)JEwRg!r!}>erpUdbc2dQzV}iPQIrD9@!s=raF90=!T3e z{9INTS!6++YkzpK*h#1_>db8Myx6CM#!6#}mfk}+q;!x}f`!pSq0hj9PjN%K?ly|1@ecBQ(n;>P5rnP~@&XSDO?e~R_-M5L)*h1qkF)LR4B=Uh{{ zk|QpmM}~&96dF~Xu`+!Nsq)0}MWC--ihI?E&ddhBptIenYVLw_s>K@ZFl=tu86cg0 zIkH$y$)33sgddM)!`x{3*rPF8@7pq;*o&1hrvG*)hI3RDfP)zRsBksy0IM)W3eLnb zScSoTyD%7efF9Md@3zNOmvYfhY1a)A{wR;RRzptB2 ztC@&uC9U=LHF=n7TeYc=-O?WrI0!oQul|xs>HE1b%K+w5M4`GT;oS@tJ7c!wPVu5J zDQ*b!13J12CY2NHV)(U;n{H8Iz9U!j!6<8cZl~op9;w>k=wD}VUeiW-CEF@hBfL-* zdm+KVzUUZxc}-J&;3arkJexw9>|%9YVfFZs(*fu#7+msFDD>n;ms{#-_TaFIdmn(C z@3Rna$tV5+KeVzwYAQRQ-93bCvGoUGC{e@2L zD*Q`Ltwl+N;vl$U-2Q$wWLVYf#75h@R##isA_YvLeJw}aJQ~K`pfjey} zicF#EvsMAt%sPgRY*CR57YQjgWbfPA4uJ#%=zST8b9uW`TS0Nn*50bE2%!LliQaRq zzWbP}UuJ>AVxT>xNC(w47&olxb)ueh+;&refwoBvo>E8_zb``ED17JNoh%%(Iq$=W zPVUg5To(>+xEzwdG}gXqdi}kx{0D{;uGGisl5?8(*TOmb1x6#*5UvWqnUZPeQDQ@U z>_4`#yw#hxP&%xxh-(t6qt)K@p^q@s9p9BqPt=y!e#)s;FK%{tZsTg9x%?Si;qtLz z{dGT9$o6AjHRDQ``Sr9d!h`5Npr$<;VL zZt)?&#N6}M^GG}2B1bTF zq`3$p4v1{r>%dLQp_c1ZG-}*Db9gcMh{%if39YDFk4SBls*6cPAGQ|(8`#4#j{nBH zC;0wM3_kY}C!~DR`-mpg=?-PC#G(~k8}Hw1>Hl#|yMnNs7r{-+tKmT90-D#jXusTW zwk6Lwl92(JW}G#77#&+MIQ{W32A}*XTE6z1lFB_5$Z(x<M-clO$9kRxTpbppuIXOPwO_Dl)rWHeQOyNYhl}jdmj;7{8eGoNCs=XJ z0aGV$x(w0$?Z^sV`I_Kb=Arv>O}1YSgl=tJ4$1B5-X3%>9mz9hUX+IL9KsxutQ~;4 zR|_TO*7A8xBKU8#Z{dd&5Hcgz*@hyrZlWS|=Q_e|aU9zN{ENck7_EEk-q(>5YQju% zxk~vPhsV7emvdfyy1oEdtN^Gl^DI-$h&*|CaWc!!8P)!*C!{yU2yitVOpO}MsfpST zodp5az=S;cek`+Be`z^6x8J}vTSif+HUvl+6%>|I-ob%p5TCuWZFD9xYP#rG$dfJ8 z3o32>!gZ?g^QEgd^tJ)e2KO(#9%lLk?P)D^4v`shZK}2MNh%@CJpzJzkv!Z;QO%sV zsMka%&4DZC&xY5Yy!UHPcIFzke(qT_o_=!0v&j%{@-JMc8M<8w3wsr8<=a>ZZK?Bkeq(}>(bRhx~YKS0BkRsBgLi@5aw}es6yNzL_`k-puco-MxFx?!D*i?rD1?^>x*mPjR23p`l^E zr*Y>#4Gn|@+%yajpk|zW%m-xa`r1Z!ft-d$o`%MrMqZvq9!TS8?1{j$JdHiIf_eaj zIC*<}d3zuw%EtlMnEl@d_J5H802N3*;(*gW4oHdi01gqjPGUtxUdo8+b;?n)`a>@&dEM!PqS+w{Gts{?8@TpQY+--DV(X*lpc+WKNQA*B;ip1ll_0g zfBp3B66n8Wrf6|R<` zxsGM(ufX(3I66B}vS-n;&I+{$6f=y$epPCnAOHM$@nwhmL}&EBV|HB%?*TZiR2 ze&M7((;Qq)X9^@$cB`oSt%l=3eE*qWJNuPKhL5R;roX}sFF#M*zFV`2tG#ZnN3%Ho zY&n(`PSPWCJ#}HN3)r@pHa!ip)1W&;__+>^2AxN#Ai#9c0%(m8N(Ba#gi-HI5}VTL zOatqr1uSfQz+q~xvsZ-3)lJKN^V(M{ifS4T`g|Z(nWqss`OvE+)}CqI$(a=LdzHCQ zSTHik8M1nC@8+Y`Jv!Pe6D;|%H5D?jCIbQV`rX~5{gdAZW^N7yd}=S!6Ce!f&QPK9 zRBVy@C6jFE9*Za9i;aKFgkyj!wBFvVLu*nZWD$U)s{=N5&~Y@-=z&|Gv=Y9e8u$1- zsv-}L&O6dH=Vyek=F^Q+q&zr5O-eUG$+uE>j$j$LNoZEk6=76GI4+SLF|J1X&Hx`w z8JE!}Z8N}=;2p_086i~4{o(bSiq&_gN`9C@VTLb3JS)k$k2t_#mj`eMSfsED36U#8 zUJSV+V15n2RwXG3{Z7UptZ<=o1E?*@p{SBwazWfo6Hw zS2y!sN)+iOPTiT0fUG~9w@@Kv8!=R@i9(p6iki*1%2^e0F#km`|Ed3;<{wTz0w+b` z)bo90BXDy<H4tizx4 z0&3aT5$yjh)xS}FCJsj*j0?|ls)H)BL9bfW*o=#yU?30^7yWpR^(4hc5H(#2@J(b9 z0%A&L+MS(+;9Xk7aG}EFr{)w-jOkIz){nKpa8Z;FRMG7hHYpE!)d22NY66gVX{alA zL8;=B4`pC~9a^~L7YL2SwO*Ri(_hzuu=*Z&%O4%8WC_p_cXMX+qvf-w3NiXB-T%}{2)Eg}_i7Bb5Z zW=eXj7C)kz6I2;0K384YrIM<%hY`qVk5A{CSS*a$oRvlEug;P$w>~e*uMVzXTX-?~ z220tzG!`l_i}8Hx#$2$sNPlHo#;Gkm@-dUix^Q0EnU43464^f%LQmceG{f-LD{IO@ zC9;Z9HinhB1Q)J-EOOD{Gv3F+gFwsop}<>E!_x=s_7WFw$Rg`1JsgZJ6+)@ABRrPD zG3D>f-Bo3{GjVA`sM#|IW?>4&udg{~XeFqYizP!j&~}xVWf|ZDwcQF>pX)6fMN|9V za~SgOKYX1Q9D*}g6_sZ=Q;2Ial;h^RD>in9VDdrgpO4?hI`=Ni;+9iQ`<8y8xwfqN zFY_~wqV|?68$}&C)t&)Ede|2Z+f zoU|{8b&s1ZQP9E|oWTP*7G5JB0nkI~YVON5@U}E)!4&EuZ(x4eu?Xr}sGU1BT4Q2B zqwV~7)*M|QM)zGc|G6<8!>e(PrcXephs7UHO3BdEQyJM0mNa->aucMbpCfw>rI%j~Kn zjuN+0G23>|nUN*0#|i+PmCM})=$P7dX9^gj9Blc}B1z|mhD6&8P+(+e5B8^L z7OO1RXih@H790V}4sK_h@CwWr4_wc)AgPS_<(pzp8K<@Daru6~(!102#dOe|KET7H zafD~svDWBLYs~w!w~te}s$}MH+4a8;?lmYue}B*<1?=9)wT2687%femWm#O!CCPPo z=GpUPt3+^&gfY2E{7vHi^qm+tcElE1NWFvs?#EDJ0);7!XAop=R=;J0GoPzfu!)*h z=RvQ076K#H=2vxx-F6M=ctyT9Sf!wTufDhzE--hW2ih^hjcDwjU>svn{+=t%Sh1G( z9E5m=X(TqqdPwR*ZpqnX-dmLP!ZZ#7dJCi$l2y$?V` zNw|O$NfZhb)&{Cq*0W)g`KL;ucYp9d>0EI-%!q4SJVKS&jMVO|t$;qx#Do;U_+$1Bs+dwJ$jJ0Np#aY)t zs(E0fH_K3JxjQ+z9M;ab-ESqCQtuOo%PQLGgeCXPZ#FJ?H14dnC*y2XCz@dmoL(v7 zCcchJ(T#|1E*7g{oeeY-l95l&<$fYM?(*r^ItC(SKE=p^wr7(7_EBz<|q`rHLU zl+L&aZJ-cJXN0#)8xgt6P7nbj@eHOMplm&8VI;06Z=)HOFyj_3paQTvda~cLd#ZQi zbOd+PYx#~YZh({psZ8QX``5R91$ z-g)7h^ZE3Du4i<4oQ%}v`MBKp1M*Bev-PQ8j{we4tkA)O%!bzDt|f}ubZIMv>iv;8`@qAaHj;8ab1y5bcyrNO4uu21~h%XP2u+_D?gzy zE&_e!pRQ-~IklJXE_Lt-ihUK9jO?Lzgg8vl^A%c)iG5HR5Ba zG8cMEk2d0gsP~Du)gSR7<1vQLy?}Yq?bU#i3W&K8(;hChm__$vx@M)x!PnBcWC?3- zS3kfcwrd~IaHbEw0jPeJrlf-SidsiTP?OxwAgNqTAjFcf!oF+67z&d~Ifm7UEE#?O zuy-0^VdEj;)$jpVaBb`IsF)iEg8ZUthD2W!14+JWtz~p_PBpz2 zj=4y-!qaveuT*x|$&8pNomKPnaRA#Rf(SAE=rD_4uvJ4ws{I)WJ@H*+}`Fx z6U|d`=M4tC3jGKqPHE+qh>%^=OdSmW&vnr?c|T(ljj=h(WsYvVsDFYZ)!|HAp7<3r?n z`S%dR^=p&28#%W>TSxG>BNtXP5kB1i->?#Z!7DmBZG$5-U`lQUy zbRDABFxA1>Ud7H9>L{uG2|ix^Teyk(;q;7~`hqZLQbzI1>QfZ~xl$lIIFcPL2Z(2V zm@@fY|6APi+i*3~0G1IBkxarpi3M-j-T<96pM$ju&@)%}_XEYlwt^7ebOAkT@Pk9g1szsz&B2^Su0TPf({t+_sd&M67GEcMve7D|7*>Bsj4U zZ|3_-VKZ}BVA6^-G!zsdJB>IU_N3NFLa+VGxGz0E_+P||Yn!SP0X6ueJmbF(w^_#LKux~mb~cTsMKfwtSAnM8Jq_Fjb`DS*pz1TE)7nKPFB=- z&seO5o+e$A#I&Cp2APo@#-+YEZKnbR6nOS4eW!Y;9^<7gPtPPP1KCY)h19cAfG@a* z6(6FHd2gq=Q$k1d;!VLxa#v#K8$3=fRf2u?{@94pnls4CdEX(yLkjVzq60xn8^(9E z!IyU)qTBToaqtnBU3&4T^ae-S==x^RL%TsAla|2hu2~nbAw9j1tu2!%iGR|K5^(TJxyp+ zho&1_j4JOt1uC@iVBhhBAc2=8R?6$^-s)P{xEwA~=5Gui*F_vH>tSKX`o3&{s!1^) z4f#sR_9>`t%`k;VKb5Q~15D?8Wz#w&74>NGm6KoU$!92VgnB$e+&+0HEgOAt+?*tO zB}T}`LR(^GT65RB7Ro#MDh?-ky#;^^Hw$b0Sr>uCtchrn;OtCV{)E%3QYgb%dIpr8 zZ@urf#kTBqUgU*_Ml*XCEl&Nj=Rl{ixZhwQzOxp@QZDg zWz4Eez)%71DqgXD^Gp~&Yuvkx99}le8%}(LP~UZccTo%1n4INCsw!GQp<;C98lmEjYJ5Gv#d|;bp6DK z`r2-2H2alp&+QwdeEsjIeV1ejd0v&kYz< zhs^*+ag`F+Tb!gV&%W7luOoJ6r|B16`Z)8_w~jv0i;>~3fgpxcDOkz-+ea7hAtw_@ z=j18`&LfNLN8Ia0tD66=Rfz-qOkmrSt_4R$MG%DBS$B5byKe2nyVSme++MBeLA9cX)1BhL%aJ& zxA5kI-t0T;==4LJj?!6`cdmP}s|)@I7rT!6uK|?(GZ}f81h0J3ezFmDL|uvc7|$jLOnDNMq&RdjDzNr@? zsCP!BT$y|BkzRAvfszmqWDXbdfWT#Gu-ieAHw6I`HrGPM6#YH~P*UCZuaV~mu!Nn7 zpks5@k(nwu?={9=V9CDg0$|T^lzX$ znrE*bwx!Lsr!{RSbB4PM5dJKH|1n%f8K=bCby0&}j%;$GoFIrKpf|>@=TGE*Nf>2E zuodV!V``3jC+@88ztmXzKCx}g0GL8+OTstaHln))N^ z&<>J0xs1YTCnB`mTzov=oB=dg;C5Q~^l$l~C#B4_iaK4@Fic3Oc@GpeY%&u0EYl*) z3o|@>JXvYc6l{GO(H-i+_=WuC_tfFL+ILNKTvD^1uYz9VKnI!^JHG4$vkd6XUP^C# zhB>2n3Iy4CP>l2|waSpLQbteLw4*ltH>{yBQWM+Yu^HC@M&DQ6b%)Nx5TTEh~>%k%@6|IY*o5W;}q=`@3nt`lp?-f`%Ag zz+E+`b)Czn0Y6FQs!I5Gy$Pu&18(5EEXAVVfVYZ`zn==1Uh8A+J0fj`Zm^yP$O2r} z;oIYm{`>P+PqJUCXE6XG?kYJJyAJs;i?g$dYKmSC+ZCV6r<2*W+MQhe^?R&9_$#9? z72ECM$j>c4Kx!K1Q|05Ouw~GruLY4jL;pDOs_5Pd8`~r|{pzGVDYp5@uR`<^zi|YX ztK^=?*8KYPqs=s<_<7n96oE2hj(Rtxu6(>-INg(d$Im74O;HyK3Y&d&)534+PODo; zoO$}IfdMN=FrCNX%}0h`tch=5+F8l7{Jv~Ccn4tGGv*Q9ywE9>DnTx}Z9ty<^l_pHl$IVXTLcc&zyN2v%C$Z_<}{m%$L3xy z$@wb|{-OQ-VAz@iV4*P6?h>{c1cK=PlA{NN8yhg)v=s0fk7+x6dD)zOofoiU8(`wL z5lz(mWO;az_~|OZ%;QVZne1>a3}%EABx;bD$z|08e>MbaDwp!DvsM|X6?ZG2u#g#J zN1!Gidrd*UranwNk0KH0oNgFA+6XrY2G`YzpdxPQ%{zWYUk}d#tdCg8rPEx5W>H9l z+TTqcQ*TVlYR##->jBch(&Y(sbZlxWT+39FspE>mJ=b}=Q=L!tlJN3jRcX>Q&?oWI zA%|c^&-SZ8H}0hQ8jFN+o<_9B3676<{s7;F2I!FD#H%l~T{;KqlgKf*?nWi0fx}q*BQ#o=DvJsfe{#cSv#WWfY_)iy%AVPJ)Mf&VM|x zHKURMBPMv9PGfDnTW3wW`r1I))LYR!GztzK=UK`Y$dUffw#$7d^XX$8vy##~_0=gJ zr)mG{zrI($h@c>ap0g3H+@LHcX3hkKh2LDkZ2nx{uZ>@20c`frl8q8( zP}AdIQS20Z@Z~_|M>xmh_bO!W2Z(Py&+^RD1KVeEbyxqH00@`l z6|Rv1_FV*9eBCdATT~)yWO`Td|Lzn$B<<(jKaD`+J0Zr(Cuz9n2c}Fu)TpP`W|4z4 z8v@N5jR{orHu7`2M>Q=a(WDMUDzrsr6GFm1!HL_XupSMG-Gl~`FyyQ^h{$; zl6T>AP=f`vq%_RL%rL(J;I8(Zl$DZvoE(OzdBl?R5QWoqYE7V9^auyL zA|zgxmsl6y8n}I28n*l1=U{(s*^qBm!sY*EKyaoQ?)v|BwuJxP*ZM!oxFB`D%f6Eh z{(Y>~ibwMlW;NFnV(M5qD~}>^P+uSd?_{**SAE|4;3a!?O z{G+|9`9B(P3x6cg8@=l@R(`*5T7TPD;;-WLTv@*=>R@+YZF42UhMG0L*pfKI_frl( z@CST7et*?Nf*iW0u4?95mYZpQhw+Optk~@=9%)Vt+>(FTF6|MTz#G8u9Jh z%4en3s}!ytrZi^-U%*V2Bq@uY7!6|&T%Ssi%?|Z!*IAD<8`Lc`{mn?Z=ZQgM+Es5^ zvGRUr{MBgNjkff@P1w|wmhZlQJ>8o-G_S#Q(l9h$z7wVybbHCYV`I$Wl_!Y4-L59} zdKb}W+mg?B61$K4x1SxJO!cVh=BzlI%v-}`tv{3^@M%~J*Yx=D;Y z=}f!V1MNbBHt{Kb-PKL4)?cPnog!AIh-W2M`Cy5znqz!U(?Og5+=PL6`>8`rh5IhK zJaHoHyK-;ab(;`RP?yJAVr3W4bXL-{e`y}P%KhV5-vm3|M9hc}#`{%HE|$sNCxo@H zeJhtSst~gb3!nCmF>&g+dlmRoBANOx0MYu*0~Wd(lfCcJ7Nk%rho!-#uLIMJ!>bP3 zv4q;ae|Ap8-Bd6f>xy#w#7W#|IAMVnU(lG%!FQ7YkzVCm%^=4V^SM(Wol@j@p$i&nUtie(&Ff*=tilhAfe^<*#dtqjMTib>* zcrzlBp7{$GNl0jP>aWg6M9py}X4cyWU1(2d8lIHgjM1H~kn70Um~7qy{dw3hsWFFB zkXtw}-aM&4@~MAk%puhiG^XHEv~Ny$I%?t5JCt^aA%;D4A8fwDaj*dHRFF;dI09l} zWS>-XZh7B>tH7+-!Z;rHA$R7@6`uy0RiS{cDuu;+A_lu!@EEkDnmkf;DERlqlCY?A z(WTyV#+-y^tyb~cpVCRSS&czrG4;i6U9Xkw2TQ`+@yG5iIg5Ai4rx5B=a$;lQQDW1 zXBq!7<`C%*D!KKr%3HpC&yd^ppj_us9{7U;^ x?lXE}vsqgb%^ARdhyN>`|0nyu!inJ%h*+>-U9nczA@#SVdv|s36sw|w{}&i&N{|2m literal 9477 zcmeHNc|26#-=EFcw}_Folr77YvNKtuvXqJlgDhpuzPl1y6pAEULXy%TV;hVpA=%fl zOLk*l1~bp)`z){D-_PrLy`JawdhYVM=brU*&iS18=kq!DJ~lDB$jWq#2?Bwz8tCg> zgFxU!@P;tJ!I+^jsv!7SHo0DiDYxL`4On0@8ShBOZKLfjAyS92{UEUd7Q- z#Sx@Y72JPnLGC`*8|cPhxhHf^fG|^6^t|bg(k(eXD%zA6kgKo+->zT($D^vCzqRhd-a8kESZDA&^B*v!)Z z5!H_Un3+pntZk|xXl#mk+=v#(UWm*Oa~z5MHa*glvSs3MSaNQ9=>wt8iZZ(Q5v>M= zAB&~C{kxPh+RbK61dwih$NZ5;x1WEm1%JRd?{rm#eIZ?f^&r{O%udkj6c08t z$|Z}vg6G)GqscG^RJ$fv3cnoq9isomp6`DzNv~V(vymm>*6Ez+oP31z7aQ-YcC(DFvEmC;g<-I0 z)lz7c^@r$RG9D(l&dK;E^-p&rY?bEVY}6xR)jPsga#QZe-4w~ibIBmXO4>xaC#IUb zF^552=F*C{%pgx`$Fck1z%nfe2o+)+uMiEvtV0F$A&lkY%ZS(~of&S8&=xJ?O!_rq z`sQi3(39A8BST^#Z+;mAs@jy-bFb<8Px>x4PIOQV(DSeK)0sfm*T6g{TKdduIjDy& z5q*lZavoC^HNVwsiC{5hf@U6DOfJ2{4~v%U0j%UY7&y>Z@)^sQ>v}d;S&u;LxnX_) zV9Nwmy+A}q0fHjfuQ^@|uK_nv?ALdGt1$v=_r}F$Davn7&M`zCfHzTw-;h zV*z8*I)U?)DaAf|gqqO~$R6`Kj%B5zFpbt>@3Ek|b{1$H7nB|GukY4f!>puxC{LGJ@HqR4mN>8BnCa zvpjs2Fl$Yj6$_ay2+0r|>t#T>L_!-Y-iQCC;k^T3?oWXtM`mer>hBXUp#quxPZ~Bg z9_uP4+oB%JeA{>TY?k@@3nxBDAk=Oqg%#2^UA%04vf-V~Z6w;t^o)ehb=-vRal}wR z;V)ZQUkk#ixpsHP;`yJ-89g)K+b5?ZBE7_8+LYK`TmK`&rP^KH#Klhb)hgx7ThTos ztC7!^Nnh_2HWZ&??%#Ovmj04Uo%EAUsliZd3cJPdujSqO$Vgsc?ii=@u2Q_7UgQV! zrygj351WW=suS{gT{!qyq;;IJToR^dq&$Wjz_L?j-vGf+fCO-n5$K15cq==W5kuTi z(Y*dPDK*ce1v$gnS8Ts}MkQL)7^A$ivAk+cm)@TB#=laHK zgbm6w;%_*;rB~88 zn6&j{-=9h5(Z8ninr>lXdQlx=6Nma`FsG|)T>uTuen4Uj zzL#M{OkhSykUkwPXsdJd-q<;79H>1Q7I3pgrW0J=#uY(uOSj%9GecL`O-fkSw5m** z`Zp%t<}sA#HAB}1czuh*-UX(4)c(o;weU$<4eFsuG^x}4lyL(4K?0Lv`_5-Ss$!fm zfEgD3q_QfZQXm76A7MkpTKI0k{lV;cep8o;{qHuRqk?RhO`uDCDsGF8rCahdUjehn z>H}uw%95QARaK)2v_#yEK;BARucuIdr530TkgFUx%kRKs>e215%nrF>Z*)5-Me)Vx@$Iw@eHoqb4 zjm2G+_ayoTMX-gv0O$J*qkg~btA(WoU^8$yc5e2eTa573y)wT@%^m?JqQFk`Cyn_|K3b;3hN{nyWo-Facq#~cMM5Q-XEtHw?EYm{75 zxJaD2S&=#8n1h82ST}dnWj_EdsH@dSTvS52gdR{$#mhuF(QhMD`C>m{eqABro&l`= zei@GdTYjj^;O1-Y9Wu)i4%~aUID-V*ZxJZeW#Wva=VRc!gSXn+W8Ps=tRaFWnZcY3 zT7L@Gt3&k650S6`lS3M(=UA46)~8hRQ!Fj^ZG5)ITiM&#YW9u4^K(wor>Q$c03Fqp zyAhVRtaM(EeSUGVZ7sTacP02k@f6AW-NAv8h4EJsOCJGs;BmyaK+1wNTvzYr}AIZjFCX}7c z2Cjd-7e^0HSxgr2GTWvi5_{QZstRQB4KQw$pGlGn440_A;?~MEq zy1Q*odT?DV{2{WMu5c&->|FhIOo_8t%SI1M1Ku~JtWWgJ9&B5zg14m#5L^{b^>%J; zVstJRy}Mk24CGcxA%B#`_C1X*lKK)f8-=nCvCtnrvHg5MIhm%iCe}48Z+zeA^40mU zBk28in2&M18w8KM`wv?G`q~Kb_TDXJ*3$XBR{?5vCXNE51*#Qc+-g4u&y}{eTanpa zwN*U@KuI*Nv%`WMtI{D3AU36<_?LQBdu;N%xaNau%Qw zCm3uwW|%M}igN$F7%~4T)jj^3MH+ z3*%MTZx;vcYmGa9UE}!%q;6d%&M?}W*_8^6_%h!mx+lY&^EWM=M^gnpSU<^9Cd~UN zK-W@uj<`{7Rqqx>%D5le|K$>_1x7v79gJz$WPvjKUxE+5^eUS zRh*!FCNTG_b}(+628h=Gry`hsdz(H@@TM46KXF68AVd1DiFbfD?NAFC&+@Um!+&S; zo;XM>S@(q*z3(0{Pu~c-lgl8}nY4bv&0nB;~T-{YnGmBh!< z)?SkEakQW8lBEb$ftBh+`zPF?4b7%J;Ay^d>FV#eA6{|4g_c@ zYY(70%gcB8cR@s&fS(wa zbAK6^W1nZg0G+4f7}R?T8SRFEk^~^0E-VMN+OXW19nkMc>k7Jh7X z|I^5n5U1LXi#X^}ygg0}87slMVL}1}19#TuFV@)z`=HOzL*Xi3`@hu4P45SKqVm=Q zG=~oyy7-L}3H3PFtj!S3*6~GT$R#2Q|0(lU;8WGLj6vZMYvTBxZ(f?-H=Nr|a?h>6 z$f}t!Q{3$262G}~Rq6*lB*-36T$g91|6z}Vdeg&y0HC*q zU-w?i%06J!WM=xWT6MPP-$nnkr+L86tnS}-GZN69PFxseb)8j1L+#Ib?Mpw{px*bQ zU!LpAHe$kaczP{80?r_BVd`#fX)e7a&wYxVSNcVtX8qlE^F~BQbMJaVCqh889hTQ^ z@MQwM`|*PD)cfug>!URB1k*Xsqx5>G!sWH^8cBn14bZg}UePmONsNrBY*4PTs_&Dm zD0Ub)7h3N`ICrH_d&SwVs%QT`sOwKt$8v2nq(9FsrK~<{_{+}6h5nJV>{$|0r&;T^ z^_D**KtW)6uk3tY0$QvhNu2_Pg4%~1c;J;aqiGl>O#R6I zt+f^XGON*Nf9^9dXPa7&n(;P;GWcI~BTvWWr?V*V4*eBk(*8oPZ(=>_1Uf#+o*Dx^ zDP|cb3liL*X6_c9D0_Piacu0hyP-A@dwwj^(ao!+*yj3344M-iX_}CS5}(N>=)pUa zk~2*#-aIYb*HnJVogDF{}v4Su)m1To$8v?hPlEHV|82y7i;75 zHX^IIU#)VRuw~dMKc0g=BxSc-dtp2Rw911 zqaO)gGNw}17~VA@17?GFJ&NVN#P$3>k;(&^~+&e!Xwkz5pb7-0c4%5^%UJY#q zJ^yh4M_GTh*gVL)?sHb)OWqOBn2kbkIh)?zGNWa$s&T)pcWG>V_7EjM&aOCVgpDOR z4%>u=(avwjwWv)oe;hnS`AE)L)7V+(`b7QrDdVz4@(daa06y3+# zKlBa)OIp$k+JD!Q%4mrA1&UzGA7*6+RCd{C;oUyZhQqFRwzclm?jJL3?O1@;xq(*l z1?`H=f^DIkWqi)#&aP+Y4x{ojb=%#1HfIz<2+gN=4-5KmQ>3?EkEU$BZIk!M$gtU_r|%GniL>Gzzp5t`n0K4TK;2EN<$i6tJ z81};D3II2zLl5exx49O?*7->6eDLx_GT06M_*6z{k)k|0PgO)=Oc@GyDQeSu6O3Sa z6#A7Z`6}!<*c2DV^&||bb&cJQ6O99F`#L1Mp8SNmq$Qs#&Cx8_uqs6f=6GtVe?k8a z(Hn+5F=aKAsYYIjtIN!6+ixFZj3xS$Y;*Vsy_`L4A zPiWCb?ZZB6{z)@Ourz?^;M$nuXI7ow{Tzre_q|s2vejl+HD3@4PpC9le2=lFmbB*h zmF-+Ed$;?PEL?NrQJN0SF3Di=w_@^rwY$CKyo()*xvT}D4?u1&f?20+9|y}F)392X zYSI{{D?_1^??V_(1*O$IXvuQ5F(mT#%aSP|HJlfH;r*r8r{c~Nl#Q6gu*4dQ)r_ZK zI!19I#yUN>&nh~m?pw7p47pGs&`_!Gu3f2mZ#9H`yU@1o3j=CKtK#9U%}cQUa87v9R)ylrTj(Jhh2^M)rHGFhaGUc>N|2U zTyeZ^GFaJ3$Td|w z@p5g);=?jBn>TH%M>Uu1&hOdf0gJv1O6e_((;pEmHrtDPQ1vmK&6q(^iF3&BV0-qz z@aaPX75N%~$|P~i4vrX^`xTs(bW7!7MCso!Ef3`#Hrg&9Uwr7$EBj|m|ki~8i;tdn&X!$S>*HCx)ygfm(70-o>I9+v|<~T!k&H#WB3YM2LHHe!%+fO2fMz} zt6z(GNB(?wxa45hPdLQbuH@!XeIoFDRn})5azA^2Hg#>UOe!Pl1n4Rh0hpaS#VZe5 zJ#d`QsULj=#XyCB=}d$1j?wu5;udJ@kKVDd7A`RSsSP5ecULn{9@#uy((01o0briF zZY)wTY&cY|2ab~0>X{*Hgs}BijfXe8_6DLj*ClI~_h(ZNZh2DM%uV*koDcKRf+_%Q zxo3{t*PyOUPqVfOo$9<;kmDMP6vK*$7mPRWimCYXKOCC4_o?szSmR6bkEg!b|JZX~ z;Av+d)_4{#21EL|-uua>BD0U=s&sk+Zem+h9(f8xq`d)1qc6B^m_wb{Kww3B#Uh97 z4rYOY`SrXKpcVnOwBWuV0xmRB>;>Fy;!BUbgy}GfVp8C#^M~Ldi~aJA`U;#S!CS;O zHYqkn)Xe!tWe(791R`W_5A5hjoyHbU0wx9Ws%@lKytvQ0P-%;RQK8aGgBc+Niw{;I zP_?9p@j#xJg@$zt_!r5bjTUM)VyfCkJ&7e&mJj=R4=;cQo3B}iH+#3^J-288O1kOP zp_9-gtY^R!U_|7&oA9gS-nbo`0W{M~kly5^EH=0Wb};tfp1irAf1(h9iR!q0Pa1~A zQ~KLqB@d{n2ixvRfMy!$+@il_-R#wwWj-L9_v$rbnyywOi(5n&I3B_6FJPNWPZ3!! z4qxnG8ExvY=uKuyoaH99MDqC4u9h(^YDa8;TH+RW%;JmSg6qkVDNvzZ@hhd6V zI}t4X`1R^0n?=Go(6DD@c&*(|RH$a}Ut>V^gX2|45ds|xqTdnNlH*lKgbqTD>}U(Q z|Ia6emLGc{hDL{ohAA9@YS(mV@JlL#W&np_b4h~6Kl^kwX4Pv~Wv~|%7~AwJmAAZv z!cF+YoHA~Ii?HQbcFi#kiy!q#2I;e%dr{^|LoF-G?$Iqe9)IV z!=xUraKsXeO$Os7`>PeHd;wcVj3NeYna}P_N}Pk*D^b8obZ6CY{_oDILG1;CKn4!1 z2#T3q`WuV0M0Jq+C!h^+&PQhbF7cNa^!GFJfIcF295r5&z<`3E`RxSmcp$a@lq9&* zO!y9*t6o@dZQ=_cV%J9qN7)!funaX?5DWU`xSziv6Rv?W9=yIS!UODb{PUS{BiP~= zM0Fzwq*eb;ccQ}}TQeJ9iy?mV8##Ul^dM}{%)-z+eb)gx6q(%)U_&!*@(|UBj88Mi zCj0p<NANPqudycGD_u#$UO(v6~mQD?}9H+ zResv^{%=k|$VdTa{{a#7fA4|QK!&XSS|il6zqj%8-E9408Qs6{AYVonIv?QnIZK01 z)G%YOqfS377Zd0bo^x7mrJZ7;I9)Tnm>zN?=U-wo zcAh$G`MHKAxm3oQKw7fz_VCE*7VTC-lL5N({wiMjId{e2o~f-jDAeD|ZV5=~d(rBH zHcVHIqZWlTW%ii^cxW$Cdy%yEv4TQbg7jD?M^&w>*zhmdCZ`TzacN(1x$uPFw#sEc!gvFJ)DveD-p!Z?6KJ zDl(HQ+W5!wUl|e)o9vZ7)=D+$3mdP;rzboh2bHGTG~5IKl=yP+FTlX;SLJKCs21Ud zGR5oVl2%c9MYx~Qyz|gV6rSlyDhICEKX?eYU?_^sM)m$p<@=ciW*I% zbM?p>YgE{w&Qmr|$o7wz)LjsFs_4?U`owjU1D9lT)8wX?YuB8Kf4B<6rlK&(8eT6r z;|{Yv=`^aKispfD62`o51o*v+48Oa09Lrz(=|<48^2m)712K4|YhU5RhD83ArZZL0 z#!SaA@5*$E?^%wI;YRa@SxSw{EO;(fKB>*^0=n8Vc;H4Fy@S-eZ^c4L*ZBZVC;Fn| zu+Qwdlui}V@2c(dMr?Ktj`-2*6d8PY$g2%ddtF?pQhDa%zYr+iG^(ns%K3SIQ*Us{ zI{tRnP%9sAi-%NoSSwZkFzfU$)ynwEzY_jU!MuI+4VHAlqDkIP z8ZAM*xaxx5kF|9C$*obmq1CP74lW;B-K_{`io#{a-bG%PRP<=3&^>LgQ6+*0*peqnbjGwwI!gz2uk{8>)v(&Wxe%mcB(Tzx=kLI#pDKeW*3A?FE&N zK)o|NGCQcT|(vo+J=@aYVqO>L_5LJyFn5GFCtYEFd5y6bUFTfIw(T6oFBM;Gh&KQ2`qY zC=d`s_!N-=q$@1}q6kP2(gK0x-Qb-0W6wJ~Z{MEXojvCz_xqB2@2|Iezx!P~X>BGY zt|%@bARuL7e&n=(fFJ}e0Z~CP!eDY$z|ZQ*6ShaeuYdqjKma3vL<%6mJw*URQB+g} zLr4J(e~kaZATgk&ATbnV3TSl9Is#@LfV;`bDA8 zDRj_y6dtek)~(+%GvB{|-&|Dm;qBYrmoMolNIC_>Ckf#AfZ)TU^LSGX29JW_(NlPI z3J(nNvB8~3l?`75+KXY$9YX{JHfQlKm_=ypEdc=~4T~cOZE@LC{k{bHNCf2GHhVcy z<>A@>cNg^O&tK{EM*9rEbn3u48TB5h{8ej=Lr9;M{Pn@lqB3e53?p0uf2u=$6E^1_ z-5~VaXSir#&mvr)S!^?0R!>O7P4on`A%nP84|doM9uEmgUjFyQ|CUL5D1rlv(W3rM zFuAxZ_YV9^9_SZmt2O)kO1WkIv#xWffOA7a=;s7`H2$1p=q_e)s-M5Qn{U-7?713o zjLi4&=qeyq3&-ysw}mQlQ9Sx^uKn7?oKppbvaw{}pNG-JS*RjlZDu&Y1i!juX(F%3 zG5=43&>@~U7!$;4v9I}{tm|f8@{-t!9;D4paT4b=>wc?_970Hn8OXDO2pLmJO9@BY zIg?@iR}hjr4dht82Tw6nFL4}emZq59bM{AP9CK#j@j_YRY#-K_sGy)oy@|z=N=I)h zaqH`NxG7Wu%y!t}>sAP3V|*S1qe^>hJ1XV? zrJVN^3C^xLC&`XXvWMXDdzFmkS5{HptuMU^Y|03qv$^*+vCV0sf1xBT$q{2O&+!UN z%H^&;{bz&iKUNm{ZDNh9w{M~7k5lH5FUL;L$O7uOU5JEs!!be(EBJ#n6*1hXtDr>2 zxiEF6lP;ROH0&DB=?RCONhH3qfIj@h4A`hwE?Su-bV3_Gt?^$SZqg9svgUM$bw^Qn z9_<|tR;=N6vDX<2M5%mmh&J_SPcr+m6o(MhJH*Fco-j2z&+3GsK}7Zu=);T0L{j4C z6*Pf6Hg@;7s65_Q=5`>j6(ZV!pQhW=Vl$?K&0n&rk^ZINVQC2h~a)fh$*R*yu9z~_;+fOb} z(1=;GdJ9@pZ;=GU3+!U;Bn8%S_=8;1(jjsUr}@#R;)~fQZkf75yTvjh&r*}-J3F~= zE{HkSOW@RDXIg`yncQSDo`CPy9U16rHuw%F56|^8wh8mlrj}>AdRgeBX@sX4({N;@5|q z?QY|aKd@T;lfMhMjJGCpJG0sco{ng$yF`5F@O~l0_J;Kq#T@&dHwPVtX5p?NWX>#x z7@KgHYgt$0S4G^DzDsQrVaiHv5UG_`p_H@-uWh#)>c=+}t3 z(56iWo?Csr+KOYe)E^elh-}0sUmut^{~Es5prAagHh3KJMddGila$rDbBNDVfAv`0 zfVYgQG5bEcY(a?nnpnsUa)Q|t4)>yV85iM?ve-d&UWQ?#8z5&??G??waZm{~8s6Ny z^^j0$f?KR!^RguSgtwoCd?kt;-ds%-&zKAwIUhfsXiDH#Seg_^A9Y+VPw=HJoP@q; zxWQz5PD3c0wT=!POd>-lWG&k?htoBH6$ut~fH;pPNAuC5Y)|-urbfJU;^%3@L8}>L%sskb4L%X3SoK)_isEPAOPT=0ONPU=|`0ENPgf2@_cj*v?l< zJI8tllP}s?lpuK?VZ1I%AzDyP-q7hgO}y@6BH@AW=}nnH8Jv)VenHnMIJjeLf!pDxlgEr8GJ#;e=vUxjIv@lno- zM2D4lYR&`iz8ZT$oO*jlCzbe6h;gt`(6q0d^x+3S{U_CJyoIvz&YOV2Y3FaICDm=E zzmpAxp5>6X1AlmYeBK7~5NN^l1J>TJ2LmK8yM!U#Ek`$rJ3#rBlYaSWi$}p|J`ftA z+yf%w6+UP*>vjAb$%a!`?03i}-2yrTpT7HooXVZFylmsdi4Wulz#RJ7>0_@#(7LCM{YJg*u( z*N>)sof)W*1v0(5P(*fr-5IFq=T9XfPF#sg{ShCD1pX^Nw{r5U1Yo@vL~039Pl3K+u`f1mP>C=H%niked{&1LP>AILFnQ$Kd`D24Ww{Wd+eT6s#C)1 zCe)5uAEY2O+oLqsl5o`UwNR*0qh%VU$gE-teAKFbvolwDjJ zH)^`h9_#iT^UHq#BPEk*v;9|Q>`ku`J}fh8{C6_zNxLbMYWB`$?_#%ndU4w_`-~5) zlwj3@+3&~NihC~TigvHtqwc(SDm0=X-wbp6y2$IQ#Ip>eW-x=Tle0%q7sINKt4%b| z#ObIe3#!t#GMGus*AvX$-fG6otB*g*owareX&;XnY$fxiPPZ0@-QX?PHKp@nUZ&ge zR$fFJ9vDVIfCP}(Ul)xk zZtN7JFB^mM!9E08$S22fS&^5J?!IUp5JG6t{ntuCEHhnzml`IwJwFEm7@&1Y%dKb#4d&pfxiPlm7f9xd?U2iaBYn(|<&+ z#i;p|UjuPnLX^F+a$pPu6kJ+~RkzG?x*SC|*->6*X2};yyq+IxuQrra@RfjIfeA!V z{x8}r>{!#R|V&)3PM z8}zk=07`)3?%n<~ao_){CG?HEf~5g$XVN_Vm%&QR|7$9a>lEzbFA|9Cp8;`}G=dTQ zn$Ps{RXgxEZ-Xc{SK>##QVlvgklE?Jlgbk?gk31u}+VcezozWbvL+q!fK{u zJSL@b@gg>F^3>Em@|QLcnh)TnV+!3h+Ju`e6B%$6i_Asyg8MqqzCtnI~E# zy1Ip;P_nN#;M&#Qu2JnmJ1rUW*VDHrRNN!BN~e*rHY?o&A2!#-C{?2lHHBGjcMeKQ z)ByDYV^7t3NlijfvnT3#-g`=X4I;wdrcI@pSAJGhuu6IHwz&xB7#xv%+3+8ZXGKXx zmwxhPN@@Qb;yaV%uj}UfqJN{9y>nrD{OPpHF}c8VPpjLM&N-O37u1iv%M~D{4~Xcf zb^+bqjCBm{?Y`*{{zCa$`SZBAk*fLq%~-AZoap*Pfx+(g?&+9Vd*&pSl@DJ%ov!*2 z4xUguhHN41QJ2VU@y$h#EJCW3hS}2DNkaD9n88${m>C2f24$0pO|x(fWmx(U;>3QqPQ641_75~5*$>zQuF%zg zBm9-DNc*-gXxGq~olD&G3*sEZ)nN?SJ>J)<=@Oo*+_XhC%sU^#MVFOUO(BR! zqEuL5<1K3wz9Of0BrT6DQ}aO$oK>XUBbUam1Zs+HKK&0K`Eg*i z+~c}Ff^rXTBvhw4w`uG-ktq|UWpY>QNQRZKMEAfmx7$CR7{4T!h^&5OgiCNN5KTq8 z)ZMq%9hV+(KfCiyzP0`HN1cgPvEdnZ*YzHE+G&l?-=AixVE$@oJKFBUoBq{PgzpEJ zPfKBB?t>lIjZl8+XiZSM8|S!gBN+& zwP37LUPDDkI6JU|-q3$(C}!HO@o^rEJfEn9a?Va8Wi)SI4J8Xx_ugCS-uA*RPN`I8qn7nhORm)Qf%yM zco55wH!@N*ziM~tcVd~QO)@&hRloYB3X`6+qkk!loehbh+1*LbGR5m#BgY1Ds3l*D zMQ4e6d=b#-C#3x(;^mt&&eoC7F?yqEj)&BGP8Eb5eJT zx??S%1DYzj+kXx0m}PXm&6_lIHSL#Vv-4Uyq>7#_VZ-VyWuFlIeWJ}IMgvNcs=scg_3n|tT{yH*ETxhh z??s+3s$rcoYQ6iqK(*PxNZ(q~xUPBTO{>8Lwr!+owCr?hfG+zHkFG{juSmuw$1Cg-Y%#ci&0C;F5@tVrtYiX%hNzlmMkzgxt3Rv? zZwXD1>bNsG@czKqWOs##Qg(OeGYn50Q))ryBJ0&w@`@(7$`FO}6E;Cf0>6cIx6K!*T|D^!j`|i@^ z!97-Y{Ckqf*5A~maM`?*jlIZW(>Jp|(>L21lhQ2=1#<^TN`!_}1F&UjR}>oQtxnH> zS>b$4uvCpi$mX4oAN$3iMd;-LM+HrNKi8tpibmpGj%k>!RshB;YwxP%ENiVU;bE!@ zrnmK%EEH@@XMQ^}$@=E|7@ZMjKozIyYYa0ifO-O{$f~0oYo2WNhih9s;x33c#=;)f z3y$2`+h>0^xnaMLhdlP3#lzIlag+BZV;{e~;}|d9eOjf6eL&L7BuJ%>ZL_t0K(0TP zc6+M!+%dDc#c2ww*JPU)1QM?%z&8i4=(0#(jSU|6gP_PiKd1g5enCQg*dl#Bs+6qi zAMWU54ROXxDrfQYwLw8s&1vaz?}Vtu$%zA{?J@~<&Dy%+&0;vTRmDC3g9#eU>V3WO z6okqh;!>9dp-jqO0=XA9^K4GD*5Aje?YS6^pd!VGtk+#c6dfl&kt(VlPoT}Qj>#M!L z+NV+{_O!i~>=u}38Z8gk_bx%5)CQrfD}py+JqMpnGAf9$v{p-0y0D( z@o|2BNf1afOe7fz`=c>AIhhC~Vlt7IOd}@KfSg}&@7_yKPXZqQ#^1j(Fp!88AtGVK zWFR30vVgSkA`)pt8j$!zKELtNqu(<#+uGXNOH03e_%KjiO-qK+h)97-Ku#bCeE2jz ze{N`qPek%*$$T1-4^#WXRptKomk zBqTXZ-_kFYYLbpSwnVS}w>NZ_rMBZhv(L~eyD1$Ktoojjg?g2J@ z2zo+#1m#hK?*nucOKeOotT4HS>5*~jU)7>Bkb_|^1YVvPcXQy~Q}LQR04h4Q zihiYT{Tqkzsc7~Oq|9Fv$9nr@`1=@Y3IH6j@b5DdI+GzIUvW@tA$C~6;Yj>|1K1BD z7#xrD4)`&u_VUvF0=ZUYuHAdy;|2)zvxnA&>8Y3dOKToyJNHwLxs3mq!(rAGCUj$I zeN5Ejmz|E&acf^02#e?M(dKDb|3+WO!R; za;r8ZIHT9pmr10QyW-hTZ{|I;Ib;)awr;(#y}p9gPKDx>x(1%nQoGcaJ~0vzFyyWsT((n(AqIZ&xXGcQhSTY6Lnqg~ zJv}?nT8E!e)a@xY&&;m|{9;3K+{ zwtk!cetNi+to(iw10ra6yuPsM_k4{d+NTC(ycM%G0Q+eGofYX{B!(A=HL zI-wwlq$T?VJ$_4UR+o3*$lWw#gzS~hh^So2bb)Gy5$xqv;4h8~9Ph2Pc2qmAh^(4> zS9kMd0`lL$q-V98NqL?|4`y2TH2(px{BsDh@W_`{WOOLNku>}(0VVWC;UR;Y?{3wVG)3;aS&wx_0@Zz=}71$1JMuaA3d^43l z->X@PXTE$3FMqtAyALFjv15j7q23Hvx<~=jt!6~cOr=6B$awLZBtvl@NN-T4m~DKC z{K|g1)i+?|^&7)()BLcZe&$KAlI{V+8hIZm$LEJkRr%5bF>Q~t#c-wfEek-i^zs-G z@4L?fw|9JDIc52QTf~8OADHT^{D@9Le?T1U+J6t*4fTsK=tpmk`jb~u?WUQBr8AVW zxPe`H5eL|$k9DK26tHH~m~&CvGuf5zPr*uTvx}?24Y!ovBr2ETt$+3aDfVypJhy5@ zL9)@X(QpBt_VuG$p?p%?J1xu`?-q{O z_xED}ONiM(^fy*8oVk1MM>zSPrvQXr*$MkRz=~yfX`cg(kVw1fgM8e18 zJ_6nIP^0EdT36!~l}|hqh3D6_RJ;+lUuk5wLJM5KPlPxP7?O3kY8Q^GT2qR|O!hLA zF7mgAcPX=jGQB0k{p(fNb>vqX?zX}d(eH|Jvf=%2p++=o3bi#v!#vE%h@nc&<#|2x zEFirUAJU5t%a30l$H$YBb)~(F};xlO{ zAkq;bn~wsHE{7`b1pHC~@!TuzLK&`P=Lc6LeGtTaaofA8+&il3P!CXw2XD&4$jA+F zZT+V^yAzTGt}J7hzX5hMQzNz!w2}&dP7J?64v&%;m3t@!Unn+Y0ahx*i>d<8yTzh< ziKX>CB?C_9X91O>dH1OvD~(fjYUb6(LZ#XssJ$f+#xBsi&&CX9?7ep32)f^<%2x0> zF5TA^*pKk8H@;A<-6>g8m=hG8BlA-}Dgacu@=vQRNp)k_CN zm8y!${tRH3)bs$!80&%nr{4i)2$u6%B(mh1p$LUCe08?9Iq4Hnso8Y4vfW%wNEGLr zk-G#p{X<@Rw+nCjZ|GC&1;Bfq28U!W3l^e!^x(9cw&1R)+;GF~Az-aId9WWwU^cLD ztcY#JcGK6vx!ch)!eQG6fMtmG5E=y_src^xG_c#MU*jir9W@gM6na*HZRuR1%8qJv zeErp^yh62Z<+7vICxv&vrvdxT$|R68W=Iq9&W*A&?G?Q-IL0V^VSIY=Y+98aktiME zxTCM`!s6Uu(ktGRr6?Tj45fgUEpJnLar9U+dbmr&e)ZW|?{TMyZ?dTHuZZxT_p0;P z2})GHV=E`b$-WxRIC+q_y)Sm3Nf z)7zZ#q~`H&RZLyyubEu7!mZkW@3M0CqByiV7Y4XyMWWmzrYZLFxCm1FI}s z=$IV>Dv2Y)fYqrDY?l6q3nt5rxFOO}G>bb@R{|5_x;rPnG4Ef_SmyS86w;#XE#93g zskbcQK#qirb|%SdB>*v1Jln$+l*frLa;HM;s$Ds`j=O*eCSzPr1e>eSAghB+b# zhwH@aYJM+OF4G&pDoPuLkDgb2uI)dp>h@qa9f%dCE)fl;O!wR8FK^8Ie&3YRp`rdV z8hsF9aO{&pA8GV_%Y?5EqX%KW0ZqB%ThUJ1OY|Yf-cv0__lO;^L9n8oHPWYcVX{K3 ze1z%|qpzibC_|SeoH~RGBiteMh0E3yh`(etgk9~E>CgDl`HPJm(3uhimWH)X2(YCdWLt`wRjBXdDa zuIeKO)U5{uT0x)&4vIjE8G!6%mqqtLvM=V~okQ&JMi1GWN1-Sf^e% z;kpuAfY(4RF?u48^McH|0&jg@pSq{b5Uww$V$!5D-#|6=T5$|D440pOwmy~Ve1Jfd z21%v$zdi3imnPdyR|iwJs#r9qk{h$yq&SJcRe##~5!3Lshh=hl7-x7zRD; zoLGs?w#mH=+mQYNjxR%Ek4lQ~ba=`M)z1lHCi>W$+BYfV&J|YAQWe-vP4bw2R?Z1c z*@V1zi*_&-_h&8mLa5o2xXJAcS*uV>KVMnnZu`g=!B^VX+f83B+CgkpP`~CGl`Gzy z*Im-GBRp6QNq2>db`WWG`R^Xsvy_^6E!2dXV|t;Lzy#oAkAS}hNIH3ivM(+ltA4A` z@Ihr81^(Pjii~w}Ff1(vw_qEqY|pMYS3%F`(mAowW?B z`o89*+0T4qe#_SG?E_S?%vve4pu@){s_jP}OJ=h;<QZN2&FA z&h@3EM2@VuNzZ1mIk?mIC9*$N{qtWBuwHdXj&;lyN|G)qeN$xD!L&-mNtR?+xV?ln&1W&eogc#8-y-MQQMhU@2iX$;DanD2WmMw3x{t4rPY zIPI>S9VKjLwiB&7s1)P_rn_4;3M9*=F}MVQWV0rNXD}3P#{41aQzpy%A$YY=vNlQB z9ue%agZSjCBHIjp$mf_Ogb`tAt`Yl0tcHq&9NCtN=MKz1vUsY%-V-(_ZEPaIc7@S> x2`U~dRzmqvsT8=Pl$9k@2V|g~yzGdDmnm>tnpn^8VnHC|eAx%QBmP7V z2xJM`v2F8S+@;aW;y{yxBilV5Zr6&}_cXG)d%LykVtGe;^^vYJC!QAOom85QNP_OL zyf~a#mnS^iA~=+u=p37jLb!M?UDLK0^u{~h6$HXN!?c66!BFF^4A3LJ^+dbR;=kJ( zIU4-Qw5|lTa!~2l=ZG^j4dWn#8<09au0E}{*pX*J-WCD!OuKAYK2c8>=6o*3I)lzoWze)Rnc^M{Z(dJjCV3}hSir0xmuvxYTzO-jR zq^-ww&cZnji-yC>KVAuksJ~!NP)lLjFVrEcg4U-JHI2_M3xcSBuW^xh-*~QQkRNqe z)A)CttAy=oFzwRB5V5IGTVGFZ_+;G+SzCWp{QZ``G^-$@p@z_T_ISd}WkEy-4dKmk zI3`UO%>b?Yb~^TePJp}ooSpdB9{#4&e`nEQl#vChOLTauakxciq3i};@M0$msvZro z66aVP4}R5guTku8IWlQ(9aZ_%t0 z5?f}%)C|^tmAFlOb|IYl(PCd4zXCA+>dYJDQ+|zVD8O=Fzw6@su*nEu?kAfFr_9Om zVSZ!n@>Y8Rb7x{oSe2cH>M4L_-`$v9;wYR#^m1#fz5%Y1UPNx~4Zh98Fk#x?n);)R z=%Y4^TxO#@8K7&uZJsb~Af|v%*ulPG1buUlVB_X0as_7Xmtb*Gbr$8owO1f5z4cS) zz2Nx5K*%nTR@pUQ6eSluj^Akp*`=XXb_#yGGP%qWzy!-mYeOpW-hVHIjGvAdF3<_s z6%N%ZbDtsk!bY@!69IXDqDZJOYJ3(Mw3#+A>s(^keX(q!SJMe<-DG$O<&jSKA;1|4 z4kM~{EFUSTv#2q1Q%F}w`pfgA@L%&RNW_hdE`1AT>ozg=cc)h zEpQimII2$4ImLlp0vm>$DIs%>+92p|2N$J*hWkT&8=Y>T>AW4d$zJaEG%L1tspY}F z;!8S(9`sHTPfW1KwY^t1zeT1DO}rF$GY{SEQkC5xKdEgq80)|e`=6*cz^eOZtt%t? z#MX*`a{;_`2{1=kb-96=UGVOYZl&vhSu#H*FFD7Ht(rt)c|}ac>|OBo2_B7EdEpD| zrVEhvE#ook%1r?6fb@r>*=kF+GCJM9%d&$#l?i)F$Y;=A3scUd)lo*gC)7JLWK37Td=~dJrEB1cnz$#%# zn@VQN!XE@2SZ8XamUrT8xkWyanz2;;vL*%xiQcLhVkVlQ)3YGyQ!O%)x+}ldXzFFJ zW~EMt;UAMzKcKdQQ<^^!VV$EAE!*GCl(nD5|5w%^fCEKGu$;H< z_P23Y{YeAAH!6iBvO{sr9T92<*tKQrJUt z$1ZcrCpA>UWmjFlT$eZEWBtb*wur@mC~zLzT$7iMhEtpRkoU|=6zbdX3YC>#QCG9OQvo8f5}ty zL?)>LnYVzr5%*H(q?P5%J=SdIlY#^{gKrox(Y{1KXfzH7IbxxK{_2l3;xpu+`U3;K z!DT@k(N)Jdux&|yfTa`M`@5dq&!n38Yb3ifT#aR_9!4C&nTE9Mq)pYvdBjn;tkF3& zB*D(xUYRxyW#>W-N~K@nAj@5E5nEE*D1p(ESpq$eUh!NEkhvN8<>N3nxwt{X_&{kI z=4ST&>}X3Q_?-Df{@bn>jgEJpuy;7>k%*p|07D!XKD|8G%n6#GsJNXK5_F+oeKLCs zDGu7J`STQKb@^{rUGiSb`j4S~NbnMpJz&#nUHQ!qeO=^V^6<+(#dXoge!P-!&#>xu zWs91n%}

fH-jfF!`^HqVeXl+MiHz5}YZ)Opg zm0Up8jL(&g_zX1POqgyoe$F_oMEh>npX2wsnxk;>Pm|dXlt2PTVX{B7dzX1bl7nPz z=9b(^1s*((ENlFs>m9`+McR1C-pPr7Me<}uSE>!336hVn>sn5jux)wz?8+di9O5~) zp?prTQ{hZL+07oQLL!A=|8k(OF;ge7X+6M874WQoqWM=3e5{6NzK>oHW+@PC&5qCY z-e1_+*HkKvRE`x`fgKd!#sop86;muYE>Spl6(RPr zETSqV^5-7!{ygR~%*{Oym0mTsI9j=NYh}Ftn<@Ugu>L<(`x=H9)Q?qPpF_|uu$l^Y z#x<4WsNp9jQgx3W#0;U_?B`&-VgWyfTIHVIssim+MySfG{r!`uMo%jx)OFW+?85PH(h)k%1zT?PUn?T z%vtmUsUw~3;VtxoXQa(}=Bf_YVAmW@)?dXAR;uJw8qSJ7w5p@Q=s6H`Vz#J| zH)k|b3^s;E5|YJMv?ZbD0|mvdyI# zKYdFk0DC|0>-+x{Oo1>)xF;AEiX%9;<|FAohDTf9HDN8KKB9WQv_rUsO2W3J*4V~r zDvccirha@ol73s`4M85@p3hC@0Kki4EpP@(Gki|H{|;J^0}jNC;<&j?3opsqp85El zjcT*s-}Bt0u9KD2&i*8b_eDHrcTA1L{2nQT@;*kvKHk8yMqS&X<#aOsIPr;+DpD5k zZ@eXSesJv}FaBmOTYjO2M3~m(rR?3;*K<{~@~o0OT_qaDm>G#K5qB!_Pf_kTu{Wq3 z@RNNf>=Grq_UZGr0{hM|d$NXpa1+(FJo?Q{U=GY zSNHBCNpqY>MVf1(=6Mj{xmXdYOqo^j5>AUAR##5KcdOwYip<3^-1eD1#}DDas$4O{|gq4?TzThDI~<<~>; zZRIwe=Xx*kS&zNf14493U&VPfqC=dBRt9>Jni2Ry7Zf!BtS7<0>zL7Rnt{r%C={^j z>g%s7PC2VL5$(+A#ny9`iq`4@8Cstor2VVU5%b^0Q1d39WmlvV-jy${sj5Pl>~F!O zL`_+Ym%L6c?qC*7QzA}}8xh`%3F6bKv+!_a{{VLQ^823ybzn7XqV*4G9$q&C!*A%E zeM#*1NLkHpo#<)oS^oRPrN+hJMoYHWZmUgC`b4GQ85a!G*WtI)lc(H;{vL>1v`sv9 zrxFb(4KU`H{F9F?q8<6CH-K9*C0K@6=-v&pcbBtQ13eh(N4yEVLJUKRzG6`7z2L~Y zmfEe1%^J8p)~hyYevGW-_3)hgnNF8=QLn7XU4fy2=HlssSuVzB3`%}QHWf_?B96?& zi9oIIe*`#IODj45@}}IwKV}e881Z#?w!S*FZon{%Was?W5tSSu8Xigrl$u=dI5^8p zdT{^F9>~HijQn>cL}~oAOCR-4W7hSR0b^+Cky2w0^5#T^H5~8XGvbnacsQeb6PxKb zGuq$0tR#%|6H+)ivS%lpE*VQ?JfWMvxX z(ponT>`wmj78+k4VaKL4&s=(HzM3Rn;3{}~7dP9yxTA{VC|*1p{_B`i-{>GXY-7s} zl%0)GJQDsy)fE(ZzI=fR!Vvr zVvL|9=|WepXm{H5fDV*oQQ`J3^5QjvqvAn9M;!9Smp z;dAC#l4S;wqQ|-JTay#KJ_Ac=EZue3yn(-pz~zj;JK?()J!GwhJQF*4Z*j*g#Ze2_ zdwZZ+O6{y#-WMlY{i-0J{2hYNOF+B%iicii1`KOBsmsxFu^ns#<`zKdk#T5!+Os0j}$d>PwV1}Tf}d7L01 zWH%6V2{>nVYyN>TP27jcXWPNHqBj%XzdD{JcP&lhTbpv;Wd`VOVc2P?8y`3e0zq8l z{{`67X~daHAoRpASZ7exrdaWcmN0Ua*-##(@HwLoBWCdhoehokK9W;wGcWY;DD6r! z{s;{?&6NllPcqj%Tg>wb3-nC@x8=f-?On+hxaZB-!7`snKNsWY%#UHL4fGc<1(q<=6GX>` zSzwoMXFtQ*ERJcm-~PlNo(! z-#1drr0b8S9!X-h6|cShD3fcayIE2rI(HYP7)au31- z0qYg6MIlU162Au~yV8n1Je&^ax-?|di)sy%lFkJeSMR^nT{BH^r;QDGK?8t7o*QQn zE8(%=fjYRu&Wbk+tX>l8Ci_#!Mo6ldPBMj&(-~g>hmW5VgxPu;OG-o)HLRn z#LYF>6_`Ht@%7V<&jYoM$j|(d{0K%NrC2logq&qTpzk$|8O9KG;2Ad z_wRr6eEY?g0IL~}GawL%Rp7?Ywm~2;0DMEtU|>p%NIf6?YW!l;_VwU51mX*UL_mCf zA->=z9TLF;?K#fkuPS>UCi`$Fgu zzH~5SB8LV3(OKXppU&5?A|R{?U$7{P?#rS_=yF)#ix1|5A1prjkA|h;voxAtNTi1R zd_LWmPmkcUz>7YS50;pG^Z6RSMgv;Z@aYr9!F#aYWLi98^E_}^=J<^vu@Hzu_QV?+ z79Q*kE>GL!^rcR?aj1lrK;wz)&%Q0Cbk}{iJw84{>?%(+)ha^;_!;>qFE$f{dPwYgpS9LZ;kxGR9Dkmt+Ghnvl zUUPb+IhQSrn-7KE+YrC7dNBO{OLgHPu>M)`wi+4Q`L-QV(SDK&-Bp=)4wABPFLke( z!-9+PP%dS-djLwkdT5S>W#Ln5wJYCR;)54Jaihe86`G%6`7Db9pr887lLl%8lu4FJ zo)te|C~`W@hsI{-LR~cGL4encV%Xc27?_#Ev6A^7#fgrqfoqkMRUk&D^P}OvMKMfm zJi73!_C^nTQmJdwwRMdxeOqQ;qrA8HNR_@9cCu4*;OFqkoei3i{VSy`^4?33Dh zHM!@`vO6dX4$Fs(NtwDnbsqm+rxxt3Z;J<7-X4ZPPt5$+#`oUT1bqG+7XI8^9R|20 z3Zb}DlFe+L8OEUEKG31Yh%CFeyRCJ2U#+cV^8+n>LV&`>$)|~`?qak#&!2=klxV~W z)B^R|40-r7fx2zHp~q-2rK5_KAb;%q>Y?XF6d(If^YQJNyUYU%mpU^vycanP1Nh@)hYEyo+ko6#Uv1n8f zjYuM~1nyUH=|U#uqZbIgvcS`V!2jWVw$riW7Y52ph+Z^clwVvh`s3k}qImaY2rIeVVv*`7H)4QH9=*?(d{7l>4o@-<6TSDWjm#0w!B#-cXj|S9ehi*kCQ zXclTk%gRk1F?{qx8;P0o+-RqpLnZx)17-wcnq{%omx(Q-wB^RAlz=pT+K;GcWwk>y zaeo-;L2aks-Il4kr;PlQ*8-J(24j!ZC&|rm32N3udhm#?WQ)GVrvkXx!dc2(w+qXl zCR9XKZg(zL9XKsnb311Z#S^y-1F%K9Zf|gq0L3YL{-NB^;T!Gy229FXDCj_nzDEqW zbA?N#&XgsTCi*vx+~|EX|2$m&pG=yknOsFnPZx6LmQ06M7*L2fCg|@(pgDaJq_KFA zG0H6TJ`xV)u?58V7R`Wwt;%cDJQ%iP zs#e0H7kPn{56G-#mzLfwm?Dz~(XK8mmrazl4Ic-|-(BzJZH*iM6w83jNq!WRVoq?? z&xw{C1&f2dw=0ED3|{?P9GrAr$XhN7L&d||K0YP&w2({L=~@1^n*ihSki=~;XK>we ztxx7Ic(xAM=gdAXvHL8^%=@Hioaq}^2eCJ_T;F64G3csx z-;#XqyII%8Q^M!sd9BY2&QwQFwl&DF*)wt>2$V(A$=-Rx(i8hkAAD)kJxi1ORh`Hm z??Mx-Cd~LRKh%u*Z|^#pVbUv`|bNHpzA551uRNu3OqW8xBO>?#n$koW= z6+HF;m!rgJ6!D@V<179;1U8@(iue4czI0cHtOWIu7BJhv*rod>6r<@DtGlB$T;+*r zS=_!-B*-OW?b5=|?r-H2OyiwAiMy?U>XIyPm~wNT6-YAp@ktO(Dk@^Uns+|)vDAIg z)qqNn)V>KDAC6H~<{7MqQ`YjflxI+G;!UI5b-)~1@M3F1q(amfB!+p7YN4cc5VAkf zuJ@_*OXt++4qag|Hi{9x?#HP|Wv+tOwzejfD>hjXX(l(doPb~s`~1+GHN`d8$mZUZ z$v$^v%q1uLhCEkR72{0ZF+lgE)8E3LIi}#_xK^Zvm0u?sZ9z=9NP3oSn$C}=(AsZA zM!$*nfL2dBZo@KBRMxim?9ev+OHH)SZwgk4EL0!h<&)E)6H~MewjNN*4#smT+|cMd zmfyuno;6@SFglMs+)$r!*UX#z4`ebhcE>RJpbox?41d{+;i>`&qiysom@QsC*<_8q zRD)M{_re9u*}b8^=&8ho9CY=QEzD^=a!9x%-eek&^n*My_D&uF#npu5jU#AjrEP1M z-g`@nF9-|2epeg!R4@m)aWklnxzN3r2cMN-!sZN z=l=}i%v}#4IR1p?$MkNH8wkj7v+zrG-2W3*sEMyDl;Kn@n&F5?25R|8WG!P`So&D$ ziNLU=0VA$bwrg0z04`ST+pLtSnP8Z=QxBG>ey2u`JQEQ8m8al$9g}qXmj4U^^Q6pq zGkSq2P3KlOFfOWr8|>jL)NEa;r~cOC;tJ`9n*jYDbqo3CBIBfH3(&9ImL3)U+Jkve z$yjq4srPlylJiAqb6fl;x&gTrq>^QVN+<@(GjJ8bXW#F(vj%sOtjzJpRP8p$h=AgV zE94K3Iv3!*`zT&IV@>lVRrYRS)9K){38}Q>`XQsmB3@0VqXaN`@H^J~22?YoY7;Oj zacO09t4aDL2(Xv1bMve>yR}|v&ope;MO5r&g@a7dfTMiCNR|^o38|y`C8Es01ZDa# z28-?*Wmtvi=X$FN>A4_;EhH*#@DW7@VjyC>p4f^XpOmFWx9NCwjwGU>vehAj+CJ&g zZI=2SKOg}?U@CzoUBUE#>v{$7hynL+B{-K5J;9J%%b>a1*thYaRxDw(v2W>hol@9y z#_qk2cn1m$>hiCZO9z1g3NcpwxEqNspAR06Xhm?2g}f1=aAYI1tSBXv-YTjA&mrIu zo)3}smT8N3GLVJg;1oGDad29q*Q{Rw0RnZO`v4oBkz#I0-3sJlMj3ACp3-jYQ{xsL zZfGBgE`;WqpcjjF=(s=%#aM_j)l(&=pQ;+F^{ly_CAd1=yVAd9U{rmEnyuTuwwj=# zM%zLj42We1Kfqh)`dVb+bJjJlY7{)Z#oOxAN`pQYY6y3>^>x#1jr7j(zi(iDYZ$Pip$0uI_u9@Od>rN(z!TGp)O4AAWHLdSph%eb7LzoxBqhD(Y zBDaFQEm~RggMo~$9M|)}EltCcpc3nrM_Q-1rmfX${;iy)9k1Sp9QxStul4T#Q{MTX zC%9T2_|#-ve7N{QdFNE55f0^EmM#~MNw_YR@MMAMIi5ED!3;Uie*z=*UBB2hs3BS&H#&aS;lLd{Srf@ruxGl-(gS z(s-AzW;=~V9fa<+-$33S0(qHw4qI;xVS5bmYm5N*)8jYeA$KfXxcSQl$7{V$XM>JPI`Rq0AW3R3KcklI!YT7RY?FpCB48O2xC z)fqH?v=d9wLVlh<*jPg=BiB>*T~pgyGKQrfa=)96b~2NfdSec`oX-!P1+sO%&`I*m z2*!kjQjrvHMy9;6A{UXPF)2)H?lwe_h3|Bz=5zDMGsrO`UnvhRogY{1IWmXPJPSX| zh~r*odzoB<7P-DME(=vPK&>xreca7lWmK>hXsY^rElhdTPVzb(JxkJ^**lYv#8kRN z?H7uA_kVgl#KSdnw?@E>jm9*Ljr6MD#7Zh(zimKh>M5$RhsmN~mA#CUtc;0~9aBIW z`dB&QI{p~XZOi4}BDXy3d^~et|F{0PXA63Mvxvw`9_;5gV!XJ@ubo~5>{dP{KPV5V zHV!>WEzGDjQg>DfPlJbnX`aZz$`fOe>3L9O=4Eni5)e7!#)%DMMU7mCGTq~_oU3jz zD~D6EGP3QlyLONFv9q9sG?qFCA(kz*!OsRq>pnhdEAnPe3qgfeVrE3z2%cuCbHyW;u=lr;pRg7 zynG~iRFDx>g7592kx6w!=%e7{5a%bJ1^C42B=x9`Brc|H-wm}rfxNf}P4HAIhMs$) z!$z?F)+HKY<2S*uJ!&gIO!PWd+{ke#vZ+5f)bi)}{R7-GRQ4A5tDU4c3k0?#P8E=K z!Uk_OxmHQ4y<@%QUmxh-($}2)D(3<Upf5U zNI=8NIcTLgD1hB}+*`Y(JwMnH8_cp2ES1~8-@1uX#Y3pWl5LK*2VT8TwU4Reuom(; ziaXR`=M08QQNZT?Yo(oHSAuDBjcP#}yHJv4^ZTHsR<_$5n3 zh3+cO2vrtLiT4w!1DuhZW6Phr<>l1@4LDV;bgZSZy3&ha8k20%=%Y`mR>qaqz7Sn-22GF zB)g;$XkP4k4N-%Qbi-Z-1}x>@1zbm9hr()cEjvkmqBFa_$j5xlpiK)v?;wIg>3&1+ z(BrHM9*oyQ@ue^he+90xP=%5x=E|3GI?R&}1nq#ePG4OX9F^=HuUMkQ+Vkq`kfo*g zMK7dx8Nr#2_QM5?U)cu>oQk6IG~%UqU%>dwM)Aw6n$@*yC*X)Uf5?$20;UMP1k~R?QuG))b$%@y3m4?8U?ic5Y^4WNfg# z(2@{)?+ik<5r_jHC~z<`HA|yH_0JdlC)}zLWn1$y7nE#lo_H`RaQ&9g3fAsT`8%`Q B#%2Hj diff --git a/docs/README-unnamed-chunk-21-1.png b/docs/README-unnamed-chunk-21-1.png index de1d44cdcd4dbf6bf65e576faff0d6f65cd693b0..328589ee91a472c1127ef23cdbc2501ab88febf3 100644 GIT binary patch literal 15747 zcmdtJcUTlj+c#R(J%Ko*f&@uM5Jeb80f}ow7nD^L2T^E62Ezy_N)ElcF6*MAtKuk% zYr+7^C=zrCWkh8V6~WPAU|0l|IErLpzT$b_?_B3S*ZIOf=RM~-P19FTb@g58j=#J5 z;4dpZb+smF0RVKDFI&7C021MUph4o2I^~EN0MPCi?=?&CPXL?&5CCUq;EbI{wwoR8v{ z3kr@F$dAg^r;j|!4qK6Z1izs!5P$-KbHP!33Pzug9`XV?KGg+k^|K=@o@IwU%f^({ z&(v!9QD=F9KwfZEjw_AU!iO6F$<^wQ*XuFL>!qJCbxZ=krj{3s z*7}&O{)owp62`>vOKM!Ry=ip9G$WSzM*?7!G5Qae@7;d_fQexF;stAB&UPp?&wngP zZ24+l?l*TAr||5uU+%nY{OW8T`tGlDjNq9m$7AjsYd?JFFFp8|>2mw-WOeQ)6T|SM zKw}(e8%Pwch+cb)krp03KBFelQ{#d_GSdzG_DdW%^vfg2Td2!M8uLuSl(!N8!-7+! zDqJ1)`{{|l{XyEQ*BXJ*h2Q^c3yR`^*Xdwe4aJ&y@OSFB$^Qkd7wtK8 z(j0Ctx&b0HgO?tDGSTSDI%B1Yz7E>z2{t%`r7nc>=n+#xAPM_E^;s!!x;%x=Lbn3k z4&R#zIGS{o2Cau{fc`_k>4KbPvsl*j@B09@sio0>dOYd>TK``K+#O;zaQfG+Va4sW zEXm)NsIvXU2WTPjX(Q^S3z}Zs_P@cbd-n30h3c3~z5_aJI@cE1T7VP=N`(;vz-cN7 z)S)ZbXp(}b%%8_reuELqrh#?eo*v0x`|n>NVBI7mx{X6GB9xzp7Lv-ud2FY)O}Z!p zMkwZKI5qA1_bq-Am?j2o1;uuwoxRkhtxNYY(WDUnYoY$2-JPNQ4bOl#ZD`0sosYC& z%E+6$yXWaRb-m8}aG3BTOEr}IgJ#MPOh;W&^~*jJLbZMS<5SSoYAk)pi&Y$)%kH#D za-!Ul0ZaKAIQ_jBUGV(sOmPr6I|ld!I1Lvuo!WUUs$q`s-zEK!0CsHDOe(j5y1nko zM+{Q+GdNq!BPdWm9{7+-TTDu|5TtRYgO`duz61}X7_#^$U~yDP`$>>i*!<`zOS$Ox z<8~mk1-2U|Y(v>T?pk1FVj2Xwdl*#V%>BSByjd58y`LbCp7aU251P$6NJq~;!a7@^ z3u3MWgNoN>mUha=mBj>K(PR(?;|+FODubC;WIdi#JU!Jc35SM8^ddRiX=j;$H2t>F zls-i$6F0IWUL2{^hzZ~UJ0=<;l-u;s_DL`7K$Qzq6_X8BQ$eN!DTT7?<#bBM&ik4R ze}#B>#Y9oim6_y93t1TItnabAU@|VWiLLxR0TdV87Xr!GK$D2pXG@7b7^qjQ88ass z9I-Wxh~EC?4Q(uOLtgu;NZ0y7VCt_!CCZy&Rjmfy2Ldl?prLF%o}Et2+#oz>+kop% z3+4O9QWNpn`nz2pn({xe+k04OCRB~#oo{GM&v)9dkJy@}gz`z^)M$V$3d>)Z8MK_5~3oVqr(TwMzCpJ|Cx=||w z(I%IVzo48lE2c8`bV?oP@v z+d1M98-{A|7J8+tZmcCStzE{K=}bkSl(3<*0;4zNm9-Aj;0#u-%z^VYc$i zyH~DstRG6v-JwfmGGew_frvd;q=Wg=DKb*F;<|0VR9keIRm@ghyznPT3-r?`azX*J z?qCa(Xgy}k9(s*-|EA9UFypjG0ta~l17D=5Wh-lTgO*c_oOB1m#@Z+%`9{_Pd{Rg2t=vLk9Qrt+wlB%+EcG?0+p^46K58Bcpi(N&X8UL@@Zxu zrgzEaw{w&a^kacMB~YKorh@jnz12~sgk*uJOU2Wv(^@fi$NP}pyP4wpc3}Ef`^#rJ zFnu0KKTLKeEB#HV`s8sOS=ev;ffwLplhT!J*Tl!)C;$3-i#v*9t)iC*Pnul_2LTp%g!azDsBe%OkK| zW7Q?P^Un)AYbjVx)cKT;o+27(LWg08>lo zjvKKcx*Dh|$)vAy?WsZ!$;ML_yle}xjLm=1_&kq^hRcDgX=MjYN;+>Mu6b{^VF<`K z7O`XxRxn+ihWAdS9CO=X)C~|5u1R}H)PWteWXYaNka~`eouT|P>9$spOUa6IvOpfK zt@=FP$C+wjx+ci;neDchnM(h!bdYOeTa%n=Ga05b(L>I!gUu(&x&~KPPRl<~*!50Q zv2v9MMNsnY@UjLnX(JmgrQf>sta1z}dpBD9AN7B?_eXUL%yf zATMli9^z(?XYh9}-_-CiWZgt9@o>K51KJp52fD>~>wWmD9|y&QmF*0407jP=7>W}= zI#9C-scZ(7uBUor3S&yr8v2bx*fs$m2Wu} zklVSdprwVV4na0r5q+UMey5)}V;Xs_K=@HW+Jw>v_)ME?WuhG#{I`V6rhhTozVaNq*wtf5kMUAzTrKYe4CpSj(1mluv-!dapl$S`Y3)LK+RC*n@#caCjY*get+= z>v;`aS5g{EH=FY0KbS@QL=Q zJY9NtCrm#`hY-|W=5nrR0Cvx&jLXOKD34`b<&iOSCF z9q#B3Ddb493BN6$==Y7-URa%E$Rc)Xi2oqkld5wJ#m|_s2X$=zN2c@+Ssn$;Vm*E}6t8WtbJXSSBwN@-`Z}gG4hb07EoN)>#G>6Myy`#Lw7(N|%upQ95fDlgrYGK$eNvle|L*1q=xqq3iqz?n zaNs0vrzT4WENJVnVPgF-G}Eb@t1Autm$SZjP~fY zX2}vrhhhWRm*m6D392st#i`7MCB-`Xn0E*an6#YscSpO&iWAMo0-34}W^@#?L@QuM zfHvVLpBVV~;CQxh#Pa~`4ndp8^44n0+<)tZ2E&f;mmr}rFN16uf)x+xnSI*m1Vh%r zcQN5&6Dq=*lEbuF1S2h|3*{lJK3Knl*`8G(d*hA^0dIOTz;MHxiqG<=$%#cyE?-JK1@13Y``_2^&+cI}P^W0=6f zFO=@N#a3=>qz4A*q2qivDt>~45if`(94XWgHN%{gzX*{PY4u4*i92j07Ae*y9)BKWZ|vaf*! zDtM@P7CN^+k24?GM~{;emf++Eg?qQUeK z^w^4yW%HkP1G&Q55!$=$Sq~0*S?{bN9XunQDBC_VU7POsz$P8~u(N$N9g2YxQpOw^ zr>hpW9s6?`+P)rG0$Y~s%<*OrA4FaSS$Wt6Fh8lwTh3K>Rz$K>q*;PfEMYoc_r2^G z77_%%W5{mSkE7f!PDC~wUNl=Dq@{v+ERfU;$6+aWEIdssk2Cj@=y+1 zZ3$Y~um!s_CqWhuGXwwG7&N#$0Q5$=DofANiUkmRMuOcyW_!^9jiXDhD^_#FhhUDS zl+8Cop=4BsA$=S~YNEV2G-sSxo|#R1PJ;px+03NyutcP}X`?aBn}`Mn1o~9>vlh@4 zz4;z%)N`0L=oT}Paw{B!sVz`m^5mnw_;1$xR4aKF7ro1kPJ49W{ znC!erC~+IBC2KtPm45Gz;$|vi#4HqJ!$g@_#U9&GkB|_&BS}c8CF^8FxuM>l2xapvE`LwTt{GI}ivW7yAg#cfuKUNRmbwsaN-A&q!R}tD zy~~LTV*>=*XOR5*8C1P9<=7MlyTk0Tn>o{j3f83cK=&eOpEsOFNV!m0_}BoO#_M!P zeH(1ssEtlBZ4}nG6X>3_XQq^(@SG_YsQMd5RN{U^alOxtDbRN*TKt+;th+sv^iyxz zL2uLLL6x9}NEbXWC)1fw-qgMoZNEMZWK^$ej-?%4Guv<$ewPlDix-6+yPLmtm zF?ORluPZ$)be##M#cp4!jl7yc&dZ?N56Jtw%pt$(h3h!_s$x2hTiJBw%Nbmm`EQMM zkK>~(sD~q|X;kFHNFZuv+kB!!T&d{49}|L^^gTn>kJT>%m^tBvbjNeOco$F}4JZdf zSQcCh_{kxfY;SjDqmA`kLAgrBJNb@bBMl`{*Z?@|H)rv$ALjB*cYNK3MG^@MH9cCB{J7((2jKN^KR><4 z6xDqkobN}{3WJ<{Ent`V_=d4~;VUzxLV;O333bZa+;1?&W>vFk#d0RDoTb6L6bUnSnGt-8!R2&K!F&d< zE9&z~-0LYbiK2#rIgodXljgpi3)?fFuwZSU6Y7ol3R|B3tX{h>i=J<$A<6{?Co!i7 zULS|z9-Snm1|VSDa8AbpSEkT!o3S`9VJc zf6ncXCg8=TZWA)I?HwRpktVqV1i-Xz8$IxVKE20RZ?E0mZeU(2J51PIW{U@`ub#HW zD#k|>h@Lj(3vC2vHF?1%w13&oPjqEPg^-q1GlEfEH1^}}a>XZ7prG>_Ny%T+1I|<{XltBG#rHZsar= zQ|4Ml9!J>!y!9+6TcWOhWZERsiocG4I`t{2y4(gG*!KfI7*N`~G{xS)Ti`y65V%qK z=HuwTw=4?BWnqMP{t16ewB^?GKM7%BUY({aCz>t$G2aZVyk(?JI&Vt|uG5F9LZ6zy zxX}9(;ZVlCHN86)JF3`pFD)%b7whHnd8My zro1IYl=>48R!epevLBmfgL;rtyW5aDZ6Ypxb8jB13tJn%AqP67>`$EZ1oA^1PrR?^ z@bVcVJ0fc4J2I;Lfi?EI?4##`i0x~kK`6FCFDjw$4%*s)CnsfLmF@I%Q`aqaue5it!5t6mJe+cL$OHqQJSjrM@CHT1pK>l$I@AxMu zm^_negx)i0HrojplEN!TaC;Mh%1>aC`V?lrkEX9XpyA&69P#aYdg6`aKDO*P6%`_xrx3;p2T<*@m{xa!}QN>w>lhApOkq>sR(TNd# z>mMI#1v!Pe%$%+_^b%}I&5sc>9cyi>o=4pvc}yq(#h2=6|83|ES2P*u4Zw`&G6pYP zn^&PN>RyXJtmYtTV)J#qhz%hy<5HnJdQ+02nux;Y_%X!wqP>Kr9c2k(uKx+_Ehx;? zQjMaZd$XxXFI&{wzr`O`{0vj$Ky=E2m8fj6Ng`<&)b2S({HSQSCm9-Skj0OD>}FuUw+pQ%ahJ`-O($?8>CLzR!(A-zv(g z&(5Nv)lYppvfvJrs7tw^HrkUg7(5i}j(V>z)uqL<7@wb$<%U0#hYZbvQOzg0w^ ze8c{GFre*In#J&}*YPdKQzWG|qK z`)VdomWt#plTgbw3=X`~*bgJmkPZgWsc0T*{q3$xNR0tCmMsh0JagJuvBWD2WO;$~ zN;2!$7cDS3F1qh?wEBClM}i@(4QoARPhh~c3M@Y1^I~;f!S=wRm4Rd-m;ab-2?1@7 zCTWWA|r;goQ=gJ^LfLqw+Cjw(NHONSiJVy+g!sORS3P#iSb2`$%3*6H)4^i zX+)umE3J5aPD2EWDgqB)Ddnd9`oX32YhSB7=p_m~MG~(ThIi+#5Fqbf_C^@m$rip# z*5C`6(pZx2SmprjyBDc99L_!9r{=s|=u!9e`h8n%`Vx>#k^QLFpoYD;t6_j(07klm zCNZelz}py9jH9{Nrc&_?pqfl2O{7+T8cGN9*v&w_d3Z`3aJ<1%4&ETe@`&Az)NWeA**cc@9E1Fa^GMqs{SUWxC7}&R zpclxDoM)d`Dfa&=01~@I;5ut2RRH+;FWQDM02_38t2%Hnyz!d`I^<6WV2`*8m;@iBzj&j?2Y@~3{0nU0 zh%<#8eiVIgm@qP3xLQ>N7SAS zWK+`^-f}v20<|8rAL;!Cx$Qp#ykh54l2v{%Bh8AIH z(Z2G)EYbkDe)$QlR$q?tpdbz#_MQZ_@{ym=$0LhC=CnmX(hzoBPn^{Ka3$hgFkiEi zKKFu$qBh5x&xyiGkm_yAu?~k5N2k|hk4b~Ps1l_Zho%D7s zRYi2!h@Hh64bfH5IT8=3V+{TlMh>?-hd$>wq%x+4#5}lS^ z>HrtYoGHTGTtx`qT^GA4_M_{YVI;Uk@`4yPjv#eElvK29fzFrme4YN+ zLF$AL9`2nWIYC`(_5Hk0egAvvh7;Ns3cPiRmXokHXq5m*mOe+rrvYHnIl31>9DBU^ zHkOmLKUu}9&4pzHHF@D>K7W37*)OSLdxK)lQ2=o)U31%l!(XE-!a-obI4W`jAN{yL zYEJ86!MfkmSvIhLAzhPp%pfUnZppwqpZXt<6TaWKoxuAlqz??|Y(%Z1Tn*k1L!QD_ z`L}y%S#o#c&}`5KWF((EJQXcI#VB5-rR;Lk7}*lB=!frCw&)fR-R7e^f3(2diz#+1D^ zXmFz9Zryr3`0f46cZf~aN}6oY#qsFdIQluPb$LRHm_(HJ{Yl6!vvtRX=*$BzwM8{d ztTo%35;?M4{rWun2D=H+d+>7PbR52W1J|(KP?n zCwy{AZc&eq>V62b6OEZg=Zn?3hCCND@t8*X-cAqYuS#Fo=2plNbb?|r^obPGuluEj?3~tTz_mE3t9o;cAF!=# zb!Z(26&%N?LJG{1;=GPU%@nxCUR?epG$D#?ou2j+l>NX=rY#m7&ICA>Wdz|g< z8UWEr_Au*rpgp{FE68M^c;~M^*+4u-P9IRLp9c-z5$#JIcseKmMwn9dwp33Q2jgYI z_vfQxCJHm@p{)T<4OkvC<6&bS>Ccw(m(bTkQ1Z7GL1h7E)GxlmB!XRID!btx-b#MZfqgI!O zA1%=%mY-KF2d-q@;ZHUwY)WKF78}Fse}R5!X2=@bH(1iiP|_8Cl}Oh^i(_%dCM#^b zsq#T)?x6Me`0O%^P))IW>~MVYaAS(PKe)#x?m*dfEp%}k`lgTk^gu?>dgxsPk_PGr z3tvq`T|L`RHdlr}ScQ%L8tjkFLVIw8KYJ5{+TA&tbKnv-&z?kH+*!Z?SFWceOl0vqjdz$;# zH4NSo(!h|BI0LXmS8OQkJ*GSd2VDy zEc2pM3^y}f_utYJKZ?_*mi|iDsK1exL+d+Xss@m5wWC%|;0?g&X&_6R=sX|K-dGON zb9#O-FlETxcpBn+#6*TzVX= z^o+}oHlWC7;wN>)FL-j{SC=gWOMb|Mvda}hV@7Z@Rku9UBnOX1wa-hyjeYI@@D)kZAhiw~uDt0UadTI8#cAi7Gs@|W;&7;LY1FN*J&GFa1;KQqIvi|yX1Se7@0MQOU7GEBRZ6s}W^K?*%X zLMbdsSeI-yOdW3eC;t_5AiAK^o}G^B{Nu;jtHk+yh;%w#P0?Zl+Tx4ztee&r^gAhJ|{3^V$9TC6e$Q z=&Hxt0mZH?o(s-gvxQYoySk&754|or7NpCoA)PL8OYe=a7gmwNJ(;9v9ZRPBywb*) z7XrO|8zw2kmd^uzat6KAPCe2&PhIk9_e=m?6e_a7p%W2Q- z2|$)J1Eg8R9H+mO1Q2!V-Ap^`1}7pk=FSI@P)(P`1(XdAw|1C7>3$Gtt|eMYcn>=^ zKHpBNp3`i(gh1aj?~iM9?5cmdgNjsQ{eh9{$3#mb?HQm!KLXWjakT#YSm=dIk~zw- zJp(}g>j97{ybyx!>hboLEq`;U80W=z=R*p|ZrhpOWdmOoic1Ya@!`sQ^mX~=?jcyD zLG}1dpt>J8Q1Oi#z~aqmy3%tUAuWJDPq&6@qem-@IoSIyzGL62Nmmdpv+15PO!PXa32%ZfyiZVlf49>qmq?`^lyB36)r4{+5M-`3 zh-g@WaqrBk-C8j*Tu?&hypkHMbLZk)gmPX*GtfL(bY^H9@d`ru~RXRk%H7mNW{8E74- ztBM&PQ8*sUJX=91zZ|<`!mDD(I5lzb4l2NDAEY}pJS%Q{d6C{f_~-_PDb<8Z0CJi$ zN1XtDV8ud%MJ#0;kSFS*1u$X~VxULBX*S3Nf$o#9S)x6gke@9`iP}$Fk5|RakJ@&X zbU2=c<&~wp5=N@_;1Xf%+a757f*Uh^stYxz%VvTb3+IdoDzSV4-E*jq?Bvw_jRO^& zzK`iO0+~^XsDOlax(qrPYpuY}YDw3vI&BmW?YIoO31km!VY?qlpDRtQ{1;x*FZUplURDuO}!w@}Dz{;;9+RuN|j|;-c_y4h}!d zdLhW(2UAXxyaw%<`;&ZF-fW_`UH%kS^cZBtbnPeu)W%Kd(NMclKg8=$N!t zVm#BelZS4sMQ_I_2Nr!`o62$Y_;u9|`VeJ`&WD%A65E4IZP=+zczc3O zT{a!fLdWQ#1z{`%5>}r|(7NdXlm~`x5H^+L!$GkXh|)HiHGH4OW-GKCudFQ;4PN-j z5(*NQqw0^UYoeT%10a`;u)>9^Ew7f-hLLzci*DcqF%OUt-d1GRJhWJs@aulgmAUPf z4bU~oi|Q|FqHQNZ@gDUTTsfuNMPGF$|H0qT=|VTe+hDE8OsO3QbKxwgT0vx1axwen z=i&U6#FOL2_EvN*CyVn}g0r>rpd{Io_KY3`;kgG*7O5L{0P{*WM`Z_ay6J$QO~Hue zcTgXA?6uJZMterX_q+;_S>MF=PAKKnyqioJg80yOvi&@MkP0%*eelvQ0V(F&quviE z!=j*b&xNr6Gl*d;Pk#Dl1$xbeQ_zL(jHL9j5wDzK1cs;pJ2_Wek!1iVwXu zI!F^n=+YJ=+Gs;pJUD`_U6BC_`vOeI0unERO0a-V3^1e1PqMJqS%G(){>bCS;*0_I z26<&qstn-qdk4>8o#c9J8hYeUsHmjW^8@g>#Eh=MT%m(Nu`!5$Iv zvLiRfIobrQv%n2JtlJ4TB;uOdk%fC{Yi7h-7E^f_s|LfCFq2ykm^hseV{r7{ZfpfI zUb5@hNS=1jiK+?)PBGVU@^uP*vOW&RTaf61mm5%ha|5p0G~T(aR9pv)1f?hTnDgBW> zG~`&Sey+m5v7Xp}${Cot9(Dx1y-$wE*JLbupi4RgsURMd`~dU`FByw+GZpG)tMAR+ zu4G=(*#DrGsS^{`b{Hgm{XVOw%Fw5NLY)RGEe8>*)zInY-Q#?_OwRd5g<7&7hNFX* z*eC~X7Bc(u;)l%o_ij8+D-vMLrp|&?BJh4s7ATABeOe{;1jU0!fDX+@KG+A_?jax}b z)<}F9-tV9vmoR(of3FreM>(A8E?I`oE3T4(CF(Qux-OP-L$&m0M%`wsQj3_Lk(orj z1|7M=9wXg^@-lFGhnsW4+FlP-wSkMt9>>UNuA2o`Ec0#|LT&Za(uP>B=4bYyh}ogQta6$v8ZLd+CyD zQZ*d};@$r*G-5&qHBe{6RMm4*nOni#@2K1?aPa0;~no+{s|I~6>8|;pqxq_PQ5RIlM^kKNrt`hwK+OeBY@ zuQg;$Bn!CWaFMigN+;;zqx}yt^$NchNg+Vvz=oOrQ%}D|u zBqU&!kiz^w{mfBA<&+s3PJ_2t4xcs`a%3#zxke{qp|0@d6Tn|bDvLBffDLhs#(CoT zh9tEy7Tvbr0VBNf@6U&PHYyCv&d{Uxf`qpJMZpT5nwcpiS{zkbZc;H3Rt=vMwmg~&subw2EA&vF06M< zY{tKD7$ZY2E&v-m!J)tY6OEU|*>6$jF<1WIKE*9uOm+ca%6Q;7;r}n!npoQ2MxzJv zR-a=Ug>dL&TC|Do|Cg>an5xvPU5DJ+N&XqCXxVsf{{uU4ic24xF+r7k zzWA%#=;ujO*4kiYJo32#dy3!i3rmvOBX$>|@*QT=B+NwCp%-A$K45e`il}Y0-t#NI z14Xa@H?OKBm4y6sVm$|C>ei8w!(Be?U$&Z*41O6Nwix-N)kwPNyrRta3E#c+U5W1_eu1;O zDzapDy>r6+9VKxB&f!Mt&qHmgR&m$ElU6_Ae}2O4*__NXUU1v2_n+(Uy`GeRD`sy< zun92Z^ru=ij1;D*i@)YwGQ9g{e^BIB9hUEw4P0`d!sl~^#f-UE$}g6_F7DAB{!zWTo3 z+WFRV@FHwV{vH zqpyuay;HwGT3C@5lGg+?-sx1efjZr+k42f;k^H@fnwz@5`2N$(zT{)-j}DZ7Ykfc5 z=W)UA{FkfCuM}1Y`Z)=`MS*VbJ}tOml)ApP%V7OLtAxliJ?zJ|FL25zT3syGPPVyPhVl%1H6Dyl7tH@t4^6$M5U1U zLzBjuN<*LaD~GP!q1mIn^J(}SUaL*K@b|~(jn`S)b9>(E1gpe`jkfC&Y%&y2^Ivfy zzIX*-+_Po#=YLMUFO6CM`oQ}eow4I=`P|N>^@l$K!=%XPp`E8MIy=26d)-`d-tT+u z^w!?hUvGczm}BzOc=k7&VvGOsZ&3fA|May}O@7|Bd*L`we*)lR`I423Z!M(u{Wk<; B_JRNa literal 16090 zcmeHud0bP;w`Wyu79fy_D5z{MttbSEqJXTnOS_@C0YMP1*y4sE3Iq^@`n7G_EefuH zBKT8UTtK*PfCi-0R#Y|>#b^jdMG)Dd2&imR^qYB~nR%a?_h$KT=5m2hRGq4OYB}HY zJ?H+heAxm6y)k+K0E0yf=dA>Q!sHw1P^6?zGhhw?v|H}EdOrCR0A~O~fU`4jCO;2? zkb?2!$J^W6lhUEGp)15W4;Bm#K1f>nAV>Nj zhe#THFgU0>=&ULTQ576ikxE0g$Tdj5RD**JInss)NevHj8pt&`*tiaU%9>526A`4@ zpsHY~RzuEU0}(SMmE{VAw{DoZjJLvCxmMaB!?sv_k*=Hq_7*fzBX44czS(Z5;?kl{rq@YDIXVASb- z|HFc!Q7EA&YCmK$zDfwCC4_?n$3Xb*SzB`F9%L261yhu^SK=pd&o2qBM_(n*9d~ zN|vJC^)mu*UYhG$Z8p?lp(DR2A)fbj*zMham*fNdeCZPoR&#(S>BJpdrRPbLSh7vv zq^pN!701_c0d$)2_qc>;?!Sx(I9+=ReM`c(`BIV#IySm|G}p#bRvG&uVm|URigVEH z+&CIog;AQ-xiI>D!N-h{ksvSf#g24v&QKrQ6nb}v{#DRtXt^`~@Zj9#px6os+CBnd z>Qv48{h&C@2=XD`0Nd>ZcoN9$xMB>Fn8U|GwDhe1Zvb~f*$5o&k#2zB*klM%6Qx;* z`hS5=ac4U@71iJ*2%y2q;x!=uBU{7z*WNl|@5Ud^5pOF$qv~(<+4AVuGtP5lVUedh zN25X>J`J@g;On2{5qHAsK!~iMHV&LLV0y-YR$bB!4eDiQ_jEQ^yI=d+2vqYhD1Nh; z-&A>O*QSTr^~;X;J}?C@MuV^^=CsSlZ*K&4!`=dVFo~;iwlR`Ly_JLVKG%~%FpOEgaOkHFK>AX-!8Hk`0jSq8oy)K_+Rq?qg40yZz*FG5i7kn-)8V)l z<^UyG9z^oTDoV3wJk{M{1LPxhlorV|aeDF^A6&Ts=;OP2#!UazNVhH3=`!vM> z_pral!EmyFX`Jod)LmK%ZBDJx3G#@zWG5f->* z?-ssh;O_f}$3b!3C|Rs01|7wMdZ<_AHe&+59Tr$k83R`x6ti@3rKOF-m;vW2zHFtPn; zPgn6VvT+te)%uqiW2jo^S#*hz*|jjijEUW^-j-saXF*9j`dUEiz(DXQx_%@m&KIHg zdi2r@u*DmH*T*}G<`v$VrfGk_XAJ1+UJK}@pty7^40RVzcMR$52*mC8!%SphgPce< zqn-@}b}af0C)8>kKU%AAXF)%Q#tj`#^ycaSZk_a<8NI@eet+NT#xGIF$^akKul@ly zN)CYXOX&SHY|m@agUJOmpr(<#8^qS?;?pBJk{4Xbb0FD3<&yLD7;rEmi&mXI1>}(@ z@F^^K=tpS|AUmU79Omt%;=1l}xSu2CFs2j1UM^0D<((jENd&rg9a@}hGG=QYFXQE2 zffgnA1GuzB%nTgYc^oR%kLDIdO#)%9*+9rn2CA1SkZE-oC!ptV207cv#yMpO6cfBT zbdL#7z%mAPE`!D}A$~A$@CVS7vRGfTfLE8tL%f`lU~G2^XmyxfqCj0u`3W9UxAHW% zx1?`gC;aeg(#CB<&BmS=lyccV9rS)U-FtThaOD-L9!qae2X=kGE*aeh`o8t|d}Om3cf-!U zReHz>w*O|tR+`6vY&UdjSJPdAHa)u%sbQi}!lR5$flc;tt}I-wL~YJ3u%bx_%Kh+} zk|`|Bwq4)8y6pqXr5|cw)a&AYdEGGhJh?x@dOkup|nywemd>wCpFimCMV23 z1`D=$adR7CdSPZcO9a3A0FTlrGSK#l5~S!ElvC1F`;FD0b{gm`5`tT!bU~fyBj^d7 zy@)T#V-@(n=ea63d$YRxZ$*P{N06IL8Alk?uefQhXu~kNM5pNG?`G&ZOVfVw&Pbtl z18oH|Zp^V*xV5;38npWChnhW&#kL6vZG7%(?XzQLdbMS+J5(_ zO>^+Y$(S2xDfVC!MDV%KRW^z)yUa_oTI`A&+K;~g{`ym3VjSu+#_GhulJZyig{<06 zr2a98)0$_YFX;|5M^p4=gNv5wXxa@g^ThVfg0xx~wW=o+ub!;AtdWrA7}nv6rKE$} zB`9(x`gVHk0pK!s3dk{nD&gfy=<^M<%*J>4%03=0SE!A4;wRg=e9Zc@FW3N zG{n2RDTm(AY%F;4bHgH7etVl02(=7&CM=5m=MH^Y@7_ioq`JrnP3{$7>uGe4boh;? zs}XD_iI?U8Q_6s&W%5TVE{`qiWBwe4bM&(pa~$Fi>RW#G}W7A_K%{*R8!5`~@0sMc$!!&PlGM z3$$dR*v>7wGKDv2-D=3ZEft0)D6^O2_9?B?=}s7JscAej4DX*|K~K+$ns96ZNW2KP z3h8*mGwDa%@EO$)WN?rk&6eatV|!$P{JG=^`H>jLkZqr!V{9;x7(m5=I+oIGAqSUC zry5P%VbSq9qhVqiQct#11Pf_J3vkL^1JaK`blp-sb1)L!bJg6Ud(Rv_{tr;B%UG}) zH{M>48+b%=ktx+G9XpEI$CXUtg<7cqqyB3qzOqaI(udWUjrS^M;hSby-u)9h_daTi z-_I)=9DaD)XhusFtmkS*d2=P79&yq2B%LBh(?ZHe1zOsWcOV`wpkw!c2GSo{Op7OO z4|iUVohNDJ&PmWTBG^D&8hllttSoUwF*fC_*6xpT{B+=t6T`4*I9jU!EoXp@quXj? zYkvEQmHPsyj%QjhCH_?gk}#Nio~P_5=rd&)qrR+wg{zv0_#Y#j=n_4NB@9il+d7P1 znuFSIh4Q(f1ry!r-sWLfliUyotO~|w=HOsMW;%-b-3EQzlQ$P_;Gp*q2fNTc=kFIU z6JXN!Tof6A&jjEx=m@CYibjH-&!AQyuI`OMl}kd(V9qL#xB*W$qI-h>GQ`CzwxY^} zRah>3>6lsrjN2wbZBYg4nuJ{~c_{sebn1Bnq_RCAP ze@I6$8NH)nIR~>)|2A#~@4{rTM(VP^?!YLQLmTj-<*r&GGAXBu zY;5W4y0XLwb7uXLb!f8%^AXgz<2pHRkGgM3g_h~ycT%3rC4Upq%){BZ@pr1qutoK(231lPG>RGvlc>$02C^Y^c43;&jk6| zJ+Sdc37gn+#XAbGfC`@N4hu$od?c$D6yMrkFDOd)^9SVzkhear)F-P-SI><~3#Muh z0fIH0t!QU`W!&Ismi%?9Ui&;O4`c&r_hheHU3y9?usJSZ0$HM3&|=J(?|nQMH=e9S z{>OrKwHw~(qq8PVwjM|nFkJdbqu1*ssB&Do0I5fGQz^cB5-YYS2{^3sx2KQmFiBix z`+A5^o3lm1lsq{)vXp}x<%`*#)8LVpVJ?se@mm`eu!JY-Mc<+_9}lOy`FM9k0Dcq@ z49lm`nakIjYeG&m-$j*SyNFFH*z^w88iJN+0tYQGnc<{%5q;fI7TYs(#+~S3V7vil zo09DmV+YcuCQKT~({&FIJv>0P|HSmi9r&6d^X(D8s23XS@b?|2390#2_UsHRWjNYc zQwY;W$vVS4@eX6gwATVl^*fPj??n)><%N!a_V`>}JqnLEHa6 z7E6D)iqvi{mss-BJR@}VC?)nte-?;8>JSA>JfZ4SGVIKeRGxhZ`G&e}thAm9FEc>zraLYdJJfLC2O={U4;*K53ccpbXniLY=(*M+j!q)lZ9xy)l? z(EOeClmW#?Oi9PBak#c768eB*MKEZofkvPu5?@s9TaK^kNc5q?|Cuf@>R12*+jVF| zu4FY|qQ78lG1{;aM_AL2`(tqM1e$~SBiWL9T#+MeG#&nEU}{JAxySp(I3 zFSe)v6xeVzm!1nG%V5hpq~81So{W7)Skx=nk%cal%qNJdwUdC}9Rc`Gy`JS55V%YT zYOVw6&yPnj>DlA;==Yr5Mo>NypXmCD6}sT2k$4YfTzov_MkEgL`W$aY6{LrT?GYLP zyK^qM@=uXK))Hky+po{pWBRjK>(TBG#mC8VVTGQ`wF9-F&Srw9adIFT8kOpoP)rGjjL$?*S~kUsN;*U2 zNVR(+@nk`9iWHXTa6*IDdg-OT$ZoZnJ`QoJ>rDwlh|a4(l~h z?)BsTI#SVA*uoVxf+bgrgyPcTHeRkV;P>{=p|o*3b<(1$rMP>=iW1aT+*saC6_gJA zGXjUmObH4LRYQqBumJ+mG}@e!xh#lz6OAgp<~m!pp!q#r+wo4Grt$4A)MZV33p61- zZ9H4^{v&Z<1o{Dy0KTN6yUn}g>ITYZEw>E6=5$4mW)6= zm#NNl&tF+H@IDZIa@O=p6M_qpLYu-})I8Mo#zU9MIFdswhaM;N(w{8R4(L#OqO($X zaHh#lU=vP$V5I#tA z)59+N_{wF4a`&wtlTNzKxeRJVwiQYm7m8Fp^Yj$({Wju^H z*h)PvZ2~$d&#mYfkICaPnH*8u(ypYfjxUbw8qq@Kn&RAAo+J+#cYl0Ca15`2pf6}V z3nZ>a-Vf2e@$@bork#SEV0JAvYXth(01?xOoF&%MCrbU&)3Vf5(= z)OIAO#E6+-0+f|aez^P2G}`B&x!7ZiyN)FCYDV(bI6QtT{ia*#jtzhPA$K{>_4XB{ zO`br{WVPO(#;LH%0@>4HO9nc&Z|o22Hc;|>TVI~iGs2Li+$I79Ms^wE;3DsppYvNE z1J_e8O)5?H5FRMSkm)}+8INZ(Zpq18MOb7Dn0kRk;QpLoZ2_ymL64Q&R(-z6^2aeO z`PIkLr(A5yEtH%_88#()%21ucU4^FD_tn!a|pDDHgK044)BbVQ!MB&gl5w#gADP@L&60g{S*Mz~)?%bEZ>dYA~rx3@h zM4Hq*5|1aGhWgL7vs(_Jd(G(jtz}r9*|F`p869p0)9e*jzP$~_Pu<40TH1qpcG3G; zoK~qhy6Qchb{@$*a{)^TigUC#>^eO)IltRn@gQWwJ_3%v5W{q*5=<8ew2WA_fvr|XO6dSwZ@l(FC-S!@u7&Zyy%QJ?yS9qqS_#Qj7VENPS zSUn=ok~zsC$KrxxRQ4|PUZ38lqujP`8q!5eTAZO`bZ9d3cpSQr5PfeuUQLmG(wa19 z4lX`G5x$~%ELKnb!pa_vf6+l!9N;7Euw^ycbxkMM>pIoA>+(pZvwVT18aVVcS>QQe zAR$Qi)KjMT3MHCAIV`B+OMuLOEUkDdkJ@a!dNN(8YYK8^Vm%aly#GbqAt>t%5)Cbz z+Ur#Nz*vKX75el{PH1;8oB1}^R_bB`j;O1FnF;eUM{cdNmPPNbTKj_!YCX|I@Z?7F#L391-T6-oHkA9o{l~qDJ=!AELuJLFS4P6Nf`Wud^JQ6#f)_#7*>! z@+A240h(e;6M*t)HogXmdIArXgbKu|Cd}I`1AJ=-oOE*ymgnXmJC}E3>E4KZ^A0(d zKK=#?o96F;SF#ur@%n{=BJ&x9@7?mMEyIHL6Vbh@D?5>Y7BJH(y73oJlA@;!v_wr$ zi=(E)g3ccR6H#btC>#98CtDzGJ-p)=Au z?3XQjD4%W@;w$zhrU+Om0wyCY7L)@#o-C0_q_ADY|C+crsP|sQd`or7<#qGVu^f6g z=rS%2kwE&}WtQYDuWlgKNZggOOed|simKho6$f#U+uD(ukcswa5hXi>zJX4$717`Y zwzl9R&okxtkn;hKYF#d$2Zxf=a0dqeT{1Mi^1BppU*|5CJyu&?GT=Q+@_N?F{tk`~A_S0x^DSeQW-ZYKxcc!qUQ;imzP6rU@yrDw54 zgfB23!ISZSYx*#Xd1WB^i!CZa-jg-mkpnBC|Bb1?eS5n?_yI6YJAw#$`#c65sRLA1uXVt|r!_BpLe5sx~^V`Q49)k}ZJ9eT4sny<-X~mF3#WH@OU-T{5gWjHn z6sb&XhF($3k0*kzSQEPp3KqnV6q~bibJSLX<$ia&NP_5>!2CqmY~bb((#8i zY^Jr==4KyyehCb;wzuEz=zsY-uFoAbu|+~qKhQNR*(q=TaSJe2C_6Nv=pkRi2F8W#v~w(A=VMMg zT+vqsyh`8l0x0`-p3XpX`8aw*K2h<7jLX?|?0aw8mSjTtN3kJv|X!i?9@ zVNOg{B}{ow$xF4HMl)vw;{4o<4+@YNO`ze-WIa@Pogk%%#T=0fjtdGG(8moKC0~*W z(+Zsc@$D{6WQn+FehJ}M#Scw+qS1^VkI>K2Wk5B-8P=IU*GJwvab>p^Cp2MA^Duhv zQ%Z6p>w*z5-eN~TNdl;FwFSMxnZEjqvG~*CES9k^kF177`4SlIe-_HaOGn|4fF`78 zgyNFN-s|;(v;xq+%XlY;nN>i}5Cd6Q5Ys&K~L6*Zvd4>ZM?j?ovpv`uRvTMTw3REH`(j}5$Njqb1pt`OAIb&5?GI*d zEeXVn=d(l#&R-*8V+^Aw*$mpR^B9+!rO6z;#Yoe*xQSKkY)-FXdzx$HC%DjOx<2E? z7kk`VsH@y{p>@fL?uB2VqPnc7n^1dKYi3Kixdkj^Tlyo+u0Y;wJOhO0Nf$+qkpqYH zEH|{ri9@(3&oY=E_5rjy8;TPy8!^XAnkb(wI%&bs*p^K776CJPBy)o0GsFqh`3{6KhGK3u@Ebh*R4C3O2sX zQ%z`~a=)$wRWB}}4e@x#Xj!M>uTCgthZ!Jl+x?VJ3Z5*?rwSD87|p0#C*mtYNkR2r zdsK12J)aQi>P%CnLSI=IprdJg2AEem$^l0px&zC1f>qP;9Ki6=xAA9)>$`<8Qsd|{ zdp0u+BK28c0|{UZhW?j@+9EfTNK}jyeQ_A$VwOmNjjy#OGiweTKZnw8q%@GQWXsPI zL~VAD>$w>O4dfrc&BnlN`F6CSuxAY(p_le5W??MM84pya*79hOZIXOT5%DPm=O^q; z7+l~n=A^d;eTH}x#YH0oX*3P%kkT>bhz{n^n41cJkLexrsXV2 z5=@j<&!=*)0H5-u`PykK^l)iRU-!B09LJ?E_>%p=2;>8w#wChFF0gRHz7SAT39=8M zp#BE+je%C9zz^U0`&E`V^K*E00y`u6Gc^eaI3gQX@wufa$n|D=%hpoOp3H=f@n37| zuOIgKwe}&1y4ylcvZNJs0M4F52ZjL2J&s5~rL4!}DNW;xjA55p#6g3TcJ_z$bbQGh z$ro3$YB#c^dDA$p#uXgN%GFpDX-PLzt$B;XQQLS>e0>D^c5jof8DZnQ3tiDSsk;$< zT_CvxN9?wQCCQ_VyAy~_^#85(=R{pHqnm}a z%Z!U7n41vKy^wV^$bDLMe31J~PMVxRibZ?GI9Qj-H5bpPMCvY8Sj=BsBbJXJOm~Y` zp=Q6%MY;Kz?XiMZPVPpKvz78xSGI9nZ%VfjUrs1V0``AN(4G%Tt5Y0G4?HUgYH;a& zZ`A4cVORbfJD6=qA2Sl)tD-{l z{CFB4F(~`+>=bXv1CJ%(m=+QwepNl4_U4hr12Ws>F#YZe)7R7Kz^Q=on`%t=jwArp zGt1=iKu3d{=G0){sdM>T3>5KLEzhM2%BRuJ?6lG~l*k@t3uL|N50>#-e-q#tor^@( zy^HS#Qz_$yPE8R~&#)$-9QbpwywHFUUZ?zHka}D@m90x(Ic0iA{@WiXppAz{+PoC0 zcI|Yhr>tj9VriZ)=7nbWSknvIE`pxaWiU;TSCO8^7O$N`@9gR^xh!mqA09mkB=+M>eW@aq0@Tfj8TmacQ4<#Czd#pN9{b4l{7J9tR98}}6x57h)Vo`dTK*L-sb4<< zKVxOr!17sGJwVZ_FWdeUw)?~%djIEmxGFY3ZYmvBc*^DY^Ch&>ag$g}>&l{noeO7a##i77LDPyfj)8Bl z_`dylJh3^iHN|EOV-3Zn+szD_6Tso8FavGlBVK6PK^<}Ra!}7odrGK=;*t+&LjbOe zWHojIW&G@S1#qyZ%yMw=#tX-W~0;3Qw14=wWjezi?1fxo7b<_07YALo6Y7)Lh*X6IGjBSvA(q3S+){*Yy zk0^x3*7}UK0ka0=N=;=C?BtdOBgF@R>KhB6W62ehbr{oNK+hc2Jr!4aY(;I#KC}%4 z=+P_T6HMZ_Rs6P8BGdhd)lHD?7A4LS=mMxoBM4U!zd)5|}TSADCG-=% z18s1kdurc~p%)aa#6hkzlH*G+k1g6lqIiM0dww>g4-I2nEt!($t-QXE{RF}P8sBTR z#KV8}G>z1Ng!Cm}L19agIc^Mju19}nWzL$s@ypvMrj;H+u5s(*kI%yS@m=kOx4PP2 zbNiPM2dP|={87-MbBRQrCv9L=6wiT1wxGawvH{a_FKbWP<9M7u@WlP=tyrxIbYygy z*wsx{Yd(zfRF- zSafXsdlX~W4OV?aDu1oT(W^VVZhu~TUZ~||yt@DS<67Bx+&+6AkDL;t#76G=ghLHZ z-FP+e)H^MdX2QZ!-|nmL3lA;52aG{6#e7;^Ty6||J;F!pgjlm;Kv@&Jv zkR$ZR4r+kz_Y(aaqod9k4U|69>C3D%J>WS7c2>R8C75_3l%@7g{xbgdag)nm#~3tz zwF{2i$fdG>fvX;(J@5Po8L%kMu1>&5H>%sap){|@B_ZZURZdX6eS>h~e0J`5!rtFs z1!|tdl>5QKc)z2*HX^?cXb#4qUdzAqqS@v`=H+fH*m%7Dk`6Kg4m*WZHwRDZu)6%J ztp3MS@0>oQE$3f zlfHK)+x6TL#&?9xtu>^?ZCp=jBM_5wGw;rj@3RmTmE3WqW5#PfeX%0M(x$a8nm3bw zcIjG1)g2Xz)2}J^K4O#ISZWVR%r=-6?Ygg$Y?f7&l*hucpQ(Y4ggD0!CXZg-0g8mf zQ2ui-vdU8Gq@7=gOA^WEcH1@LyN!Aqh?}GtiksKCE6RCkNmP5Dc=Y@^I3f#eScB^~ z&MCRgVw#GD%ho^{A^LFlQcaOfVKh+5Ns#f2|qxzS0krKNgL~che8R=xb|0 zLCN>+oi?vhQzX5rh2tQjxG)b@GWlb)KVJu-JJ%(+wBGm7W{i;ydIEti+L(G%iU}w4 zNw@#wo2ORk8{;jU-gr$gs0tt&t21=~v8gw=;QV^WofZwAZ z&>zcqp^bi?I4)x!?3LGi4e`hC%E`R?!_w~$6+)w1LRY7716g6{E~KD>Z0XPK*`F$% zaZNdlJ}Rk7I2+_1uybilBkw!>sz(g<3*Of(g(6G@*_t=eBme+B!O3gMx32f`=D=tt zi#F-iHF&&`w!1nK^K+Jx9NE*Vr@=5N^Xb4{wP|N+=HQA)w~xPvjT6I+$+Whe_< zMtH(2xWBt!FO#Yvw8|wjSZ6u~bim>(%U-e5?tt=b$nG3rRs+52zgmB3TnXfR&-i!W zhj#g=EpU6E6Z+#9P`Dt=8Glh(zE-<^-F^0yL4|%=A8IoffG8V2D>TFfSn@p2fvW5$4L#+#Uv%1;jDpHcYCc#-t)Yes3IOA;&B^HuoT3e7Z8ub1)heY-1=e9~%uo226 z=^(kPwAI;;j`Ay{J+X8$h_j{%7p%JOhKm>WP6T!ZQ2XarN~G8TiiAb}6F`|h>ThbJ z>bBiIK&I#96wrN&WY^Zcz8_4{gE1hk1u1f-9+(*mT0QcKyF$sro|U6-wfO?s;Ke2D zE9?DG(+nt^{Sc(PRO@PPpZ8@xRi*-B9&XJ6)f^0u3DOpWijMCrZ6k2WOe2a?G0c22-f&}yecHqBs))q8GTkn!JX!g;!4lDV9h&aEq*0xaat^qEMketT?) zIa2UbSxhO3$L2(0-cNH1b&Cq$s$Ay#lX#Oo`mM1+T2%hIgfO;&1EJ@J)3J?tXJB9M>r`C4N9lJYoA z^J*Ly2?WLO7lF83JB#>qYfk6@Ne5MYwe4q^xNQ+=AIa0cUFmC}^y!=X;XCY^3N(S!fSqXOm>oKrdOuLd#cERP*a%s| z!G1dvkesKB^HxI3(ME(38tKiGq#vSB3Z+JVv;iLS~LkLK=jTlPPm%{G1eNPwZPR)cL=HwF?qwpcH#hg z5Sve7Q{p`Ug4)1js?F zkj&+9SN^Ia?EhqeM9>p9cNuarEzhuzsQ4RQ_vDU zbb%8WI`K9+IPy=`4?aKl<+-pU;jv3+05-V|%mU)^0^i|gfcNF|Ll=Um-Q&-Z++8n< zKt=mkJ>zT3t@_JIAktvyK{i=xsyozf+H~@^0()qAUE^z8&V}_?#B;Dc6dQn!zy;BJ zL71Q4iZ)hXsCJC8#hh4{21S;&0nIiWn~UZs$U~8^QN|@53>6!oyF2^8#DAS70Nrd1 zTaruxuj~1dH}5vG$kUrK?L|*1>t zZY;v>Su=Th2h;Z5yvz}4k&I=kvVvNUIj zlqumRh$LZ;0A4m5+F6Y5-~T^bu@*Q$?MmQ!>F5nps2wo1+h7H>JO2JVc9ize;+MLh z*zCV+qkbNAosa9ceB$FjtT(pjv2i31Zyx#-VS~;!K`{^4JUXthQFF=TRY2{vD@{^6 zkmmxHt$G<===*4UN{JC5du8*6YV{H8SSuiovZg!VxP!biSE5)Tob=v$BX;G64mM~E zhJt|MXzPX641fd7@a-imneDNE)7(t<^)Lun@pp?PmOzMdE)F#q8hC7b=IWO&5X}Oa zKHYV07yeBTL^m(Pd;WY>ult|J+P>(&?ZA?ZZxT5#Mp%JH!ds9J6F?(~9s^7)hsJ7v z5`{z3x%eJ$s7FfkmZIB##(M|odOs-p^tU=d(bvD97zD%(^)drYb^mu|2!_2}m~0H0bCU4z!m9(0@_v1Z zb&E{g>I6r1RMG`w3)O<27h>|cHlisy{vVI6NX@V{5E9ZZNQpFjJWBlzjVCc|NE zwp@+(?XX)h_XG8F|Ga&4M#>rPs+W@Bv9CshrrbX-kC>VE=>Bl(N!1w;*iz|OU33gD z9^<)a{gf@@QA@~+dsh$NU-^s{r;wKfH%>fo_WW&HbQ;R47LQr7V)t#y92bsw&g7;4 z!*QotXj=w5zYbg)?ArWU<=6WDNYDM<5AE~*vo%fT9o%eLGSJ-LZ`uF*qfr%StJ{K- zs@+qXda`>Kk1Ai&3_a{!-eoitUVYYbPQK*}@5qLLAa{qt@7?D(`}YJ6Jg#dM{xso% z90$5zUNEIsk{OAvCU3aZK&*0KMnY-dYk!X=yXNxO_Px6B#MCoo`&#a2-&0S0JwhKh zSS$%&&{=iv%c*yU4K8$b>xxJ2DdgRd#@?5$>-}tJX*M?gGcL2!Iz@1i~@zn|_^+Bt6K{?|6NxN%ZxQ~Q_c$@k1BTCWr3oaMHLJS?{g+Zg*< z`@Pw6us6K-*KAYI+U*Cp2cs-!DzbmQ9iU9hI+`E~av%6dul1nBXNiGw8>8xT?N3#t zHw!1V6rBircXt1K$7DOlz&mv2uBN%+M@gTZy@UOCdwU1JZPs1y*G8h~>xhvraXn&w?d|?( zvfLl@<^im7B@f+EIyhxl*E3hoME^zOLD@$zi_-owMST>+P1+$ z+5Pz;4fc385Cs3ZsI{kO`u(@u)p*|ZeAAJfqTlI#FKv z?mMl!w|lvyTV=j1d%HzieDty?g5NZM}BcnQ$KTGc86EHTXE>rPr>+jLQz9_ z{GUqCq&|-ipPzg2zdxD0C3$7{<+T3kqh}i$#SD+P_@BKG{D1TU?|=BM&qiNur08`ogb{AKg<=Fq?VC#hD%TmS$7 diff --git a/docs/README-unnamed-chunk-22-1.png b/docs/README-unnamed-chunk-22-1.png index 716ffe674b5ca44544be813e01ce0eb3a7eccd5f..3fd3d6497ac132a5903db752a68822ac8ead5447 100644 GIT binary patch literal 23402 zcmb@ucUTn5*DhM!JpmYDBxf*!3c@H#2Dcj%I*17vB(+MAJW9^3Zbd~wL1D<-1Y1EE zL9#%*6$MEulGBI^l0hVgInD3;zVp|)_dNG`Zd(jv&#LO~npL&J`z~%AJ9*r27QMVwtd zw|Y41JKxcFF4Vt6v`(|t!^_Mp+RV$#%!_!MYQkHM1L=$c|}K?MQ2qlzh`-U&x-zDO|-1<#KGtHpMfSH z{XtxY7>?-7?<^#Q_}__{Og%*{$Sszy$2)tKTD>oV!+m+L8oxbbEOkSq7 zs>wf3lN$3W^7wPSIu+Y2OU-K*0p0inz z(&G=P1NmAj z^Ho>)ODk0dm7?ZDBB(ntTGycebxtTH)L7(^$vYo)yk$76vH73C3zJv)3)$p0=Ig*j zy)+TgrpDZ$RAvMNIYAH0i_Qvk1=LlsY? zo26&|Z316Z$9pR@9ei#~{^IGlZ3DSd-jPctMv$Udya|-kdv2*FVt$+s1#X=5ucrBI zrz!4CGlAmE`NGk-&5c}>hTQAZ1?PRM1EVQ(-MN*E5?GO#dE^)AzkBD)Q*MjMHa-C9 zk&r#&y){kbDIH%ZKSRihkU0^vD^*qW@Y$ec2|8|bsK~_L7O(1{b@DuL*uvL}cke=XGE^>*M-w5Kl!$Q=X>pX!w6ql0@-|bpHIU`x__sMviEL z(7*pSnUMy@$3CuKqygnEKkqD)F}j3r^~z3YA~|cIl+8sVay$%Ma zy+PpAYUdiPG2HX{Y^%9_f%Jdf$KsJ~yghct3m+*<{X?!UvbPd$lNIwSH^CK^EA!Ht zo_Iila)8vo|GDSIBSWHdy~DMZQoPQl-tde1%O}zgRH|N{%zF+dR(5{f_7Y*){T06S zBvb&mSp&v36G5*-x^ZxogG2g=)xZDsxDhper1&78C-m2T@7~d#45}RzgYgZ2Wp`h` zRYRLwk;8NpfE0wv0*&qRWLt~c96ILYfqUn?4O0k6m>E^@L7Y8K&;Qf*DEo^`T0LxU z$pE#w2YK=rM2F+zmc1sHFlh;on7rTDa*rru`vL0!6i@9fJ2I&q>$B&_CqVU}+gUI` z6yV-MIF&1-gTH@TH7tMZq$+&6zZawG{FCOsJr8ywZh?o-y^HenqG5{*d0H8^cu#x` zoDoSK>Fx!F>pRI^d#qaTQreWZn_yWhf?v%2D0q;02~=Du`otN*LS_cmqWw#IG{UgP zhz}MwT|1$a;}b8S;UGOrapQ)cM7>oty2vorOwJz{}jtRIwr>A<}tBglF0 zIXm(DSt3ZN*O4u}l}`{3Qufdg6;FKNy(|>DX2un(`hErv7Vms6OCA{!muC~d!&bp(63g90sJs-2*LDlZn6@7c;=7!| z9fi>Ozv=R|A^NA0L-=zgm4FVOkp|DkSYvY(u1Ls%Dk4zKm{%3kbO|Pj>49g{ia!W(Aa*}sd#C`ZZtW-mUZ7>Wg z^&VNq9_`ay7PjALO=FlM>W6%xroDJ@j5*ad;@2n}t2dUDEqRUE*t{g;qL-mS5&R{s zIF_GTuh9r`N-+0KDr~6hyNp#yF0R8*Hh4gR(QkpuJY>}cWowQ!Ae;lR<+r5T%3;aa zFsS9pDOECbRC*2Wvq_HolQfA+;pzemS13NKn$D3V7kgQ6LLX+r6SuZ1a{oH3PMK~G)#d=bnQFfzRFvuy(dKXJ(#08(m+MPkv4I%DR2d^i>97SP< zUoG*KNs-hSUMRDTDu$~rf{a=yBMuD5E#8LwDZ7Ayo(3;?iwLYD!W$F0g{Vu3Z4d%X zD`dpt+pVAfg4q@re{qA9J?7vAie^!Dw1vzp6k^p`Q@Jh^@d9n(;*WB}fWrOr_z-+i z8^84g$`Hf^T_I-;G%LL1{;gYZh6G0FmCsv*Ba3em;O0vogle;_Jn_CPl{X-2|J$ew z%DiV}4#y%0s@<^%aQ+E6rH@U|8-qv+HgHG^uhY<6jJSYXY(upL8SS!Y&=WB%KVuiZ zNtpe_?7DXghsq5NH$|9ocuRi$XRKbkk<0iVg59(eGU z0?mJ&RmkMshOcBl+sKHOnNE-Fwmt@+5G#xdh<#Rl&-fRZdU4&IapgLWU8)pHU zMle$bsx6Ox244NFMh1Ab4yErQG+Y9eqZTW5+aS!HAV<>{Vf)(R7eUZzS5Rj28ENZP z0`?HQeuF%kpCN-^v|Eq0w?F~OaS%7q$J-Py{MR0@pgiM2fc zkTt0L(Ou|N)3L3|G(x*z#uV)AoL&UQxnhaJEP7Ko{%ejS&*l8-bS?y{{jBA8y~ArN zfPy<)u0`!~$Q*4+s#kj`W?)Mf=A6c!BAoq*wp;>GSM9{w*P=S*SYnWr9H;|b-@6eiv2b1Hd_Kjz!Rj~C(G5VnAF&nh5accH zu}xE29%{z_dJZy_aE~+ijMZ_3?XVMe7DoLvn|<*i-*qsB)O}bCDvFr>QH^<88exL+5Hl1L&O-6W2&Wyhw#09hV96PD#4-*h%Lfp&G z+JaSHsm4ljO|Be-RgUU`qYX)ztUQ(3PpN|;dCw-O={WwS{cSzmvN=(>ZHpkxEF{#F z3d+7c$?NHEexI++YY1b$#uU}^k&IjQ!Yof28cE$@xMvq;BW|d;;WK2L>nVpiH#&LZ zU&zEGOw08_S#eLcV#8-ob(dGT6Sh6gCdQs$PnhCIuZtA3oWF>NP5Eej=zP0om$jD= zx|75qAuXU?3W`y5!>s_jAg@@cFwW@SM#V!3AUoVzISKdnV#dEH&A1ugCz z{VX_ZYnEdG4peXTW@t2|! zX12bpn-CAT#s*KX?ph*)cRHIMH4e4zXbe#e)PH;wDZU9j>f^Osj~KI46D;hEwhC-fVf0xIO#oz3&dNVHx60v8nOR*7>QSI zCkDbxu&VLW-J^w&7Q*qQFl2%4IJwXY@1oAFobC-yrg7VrgLQwy&L<(A%b$T(o<6Vt z&B28Qi;tOiq*O4qtGN_csVVB4;(g51u(o%56Bk~XJ^4chJcyxta8IU#feVs2HWdf2IS zQP6)uF?p`O`H@4BbhR)`?TinseKG_p;}@H$Ni&W>>U+A>3NOeoZcsz*t?8&3jiqr; zDjKK<2pIrg@>s}WQ_SXErQcIMHPk12l&l`SO@^E8Cc%~YXr@<0Dfe=j|E(c6xYk&f zVJFA7b7!DICLH@}^%R(;Tpb|5dgcDW>#D!^H4#uZa!@IbP`rV6cDSPmp3TJ7CAr=1 zO_18&zQMO6sm6G4?$FJW)7Ky_ZEoh223jhN`wHV1)8UErP>hNKm%o(g39$=YaJTEy z2(D}x8zl9J`Ik2#rYev!a5;*f)c~_DB(BlJA^}bVrEO8Yg|6F)B}#JV zbQHPS_vP_lE({O6FDZ}CeH9>!<-eyF2idgBegvD8v+@ap7osAC#*1(g zixTn>X9}0K1C+_aUEvq9s59A^jl~Oq|K?#;W5Z;o6r8>EgQ_ki19=>+95SNd^3-fA zH>gFcpW@L`DT20t7)7)fjl{St1?(L0z0^@z>PZm0Dc%_ORmQ6H4KTGPS5vG$Ee){j ztmP1929o1}UlgR^8^ZB^vG+`@AoA#?i(p_u*F1wgJ^^LI z8j{?O6H-fV+0cT5bpTKGca^t)6kkgVhr0>@o+y z%4Fb=$!%V}x@gC?1R+-HSpDbqMp)Ky+~YW|PU4nqy(qzbCW9RmhdA0{UEM*$U~Of^o;PoC!L^tQJfiZNCwMy_QV}kaJeyctmYk* z5AbK(G|}x%G|mj9f?)fF&tVp4Xhh8PVnl{9N>xvYdoP=1E5N%+c*}m!u+c|)Th23t za~rfQqOy2~J{U8VL;G%pAZ_{qehcNYN*gg(tw-}}vW(m0y6(b))~zRYWQtgl(8XpU zHF<6l8Q*4vjf%J^;o3X~+Z@)8!-tM8b~hg(dX_?W3S?zR;)q?kd4`*pvaJ%VL@cy&gU#`)*<-x;0?+>X-w zAYev6KLJ5gH-dQaSYVU&Y3hp|5{6kUVW!qtBIGez_7oc}-2iO9ISKI^7LjjQ{mg~Q z`SU=#@qB%bEK>I#w3rX~HR!c!X0hKpCNeFP-#LKZCla^Y7WRK_NG?wt+d+{!Mn_{L2`YiH@b|o3| z8s!+{4{1z};Ab5zfPRjPW|yqwzeF@pk(zL2lALz#2L7FwHy`5(i!`0$e@jHAsr}TSOS=VmCJ{r!l?%-*8im#`ahM;qiu$&pZ2A>bYk zdG%A{3s$oaZ`lrkl+WwC?8}w82A_OQWzak^G~o8kWuPDqD7^FdN{ep5mtF7M&g0pu zvAusk)uWC5XA7bp{<&L+35tc=^rT8d{QrKK2$DNZcg!-xiI@nMSXU^>K+BupK5rB?tg48u!@PdCp)2FcoA)`4ii z)D~v(KRc6nALbW|r=gGoKgS@Q8}LZZ7MlcS!)Z)Sm?K5wbYeE&wyt6P7{ecCLrkd6 z+HEhsBBiJHp=EPi{wCDNyHN)1Hp)kGR6xHS-ZEPx{ z;N6L7o4ea@V56=?O!o=kX%a$2D3P@Zx3y@a)IouIxbs@StdpNGv#ZP4b5r0!38UZ( z*XnLzWHT`eX9o0_V^4R>U4wi;ysJile}51tYBgpbd$OPmW?TgEW_2t`85M&$N)Y$_ zxqyrS(~ub^W~z*um-&LgO>w~e&$HKrIrYNDUW0j7ciK%XDx-G;y6pOQ z9cJ37ydQ3u-R{(se-ID#!2|~(b>rqW_!Wrv;2y*gfuwec(C1d@37H2ggMS=-Wn`p0 ze8(C6z<0&Vb8f<28hs?}U@XkZ5n;XDwGp*$251kJGu%73rl$UB!%3HpWHm7J^2mG8 zLRs>4g8|~Kz!(HM+6X5x zfh!KCC=)>!jOEeIR(kOzK}$}^oY7q;h~8KyTpYHG%6ss#v285(>W`_oih_>}bh@=4 z64%y5^~HH*wF9Ba*pFT!I-~O*bkNrn`@D0e-0>4i5aGz%5*pafF$QV9!Xpe4o-hMAq)VcsR*w25Z0Q#?3*f-twr zXbxy%;D#_GK{9VG-lBWVK|D`^cX%}a%f^Y?g?r~!@V5&^M$MJkQt1%bV~7TE{slmI~v(#7hX32Gc#Z*KFp4P411-&=uSzy9>r?e ze=qIi`5~xI3(k;Z=&j{7xF;p3-hUzDsw%=h$P73$zo+xYUsl`D9q0Zz5^t8Vksa1# zZghI3$OChwF*dh^6r3aU0xC!ZRxe{QHNoOb3`K^;BlGcm?#Z7G7F8xtOY?RTXH=9C zvE11-n80uJGvwWjgfiq|TUUJA`8!;f2`Y?I9j1EhN;B)G(*HyzI%<3+PgkWLTZeXk zcEta>4RUmP<;W$#CZL6t%2mU?mhWW?6fZR*v#;o(Aq#D%&&#mGvck7wB_1a!c&G+m zU;f;H%mi%-|E&UY(uj(d|7{_FjjjRxzl^U^xEvjX*lyn_epY%C1OjAyp(0}l#$ByxT;^tx zs(C<*(MHFpbvl${dm~M0#)XX1s(d)-cO4rK>xA%cg20xH^v7nkMG%U98ACoJeFo~HA+DGANG zXK`1WOQPeC2BEq)pfw@Z9p5*XBs) zlZtyL06w#!z0~od_L|}(h!F#22iK+%o5nL4-1h*!OJp_D%hxB&ooy9tnSMu^?9uRz z3dJRYsc*n?m?g&D?Wcmb2RjD)3v!A;D}R?GCz~OUh6mzb${vTZ)}cn-+xp;)p7OGz zg=EeYF}o7dTu%ORyk6vNT~KV}oUbz67TY3o;z6INBEoztt#Xgl)^Qkk!MNK-1lSl{ z&o62Cd=Y4S3v{)=S0W8UIashI+nKxD&-#EK-> zAi)~@5lf_n+5s_Rd0Q;h>IK_QCru+>NrU9<02txP0`YRAFQ@;ea(~<-0#VjqX4JYT zIwg?z(X^M679UjJyz zyBc+xh5?Xwo?Ya9tTzb=eblJ~LnGc92NnsnoZgb%L2k;JPqu z;AhHP4+%ytv5E(F00UDJQy3fz?UKDhv}OMj2As4*GXsK@?m9^V24CyagD_R9lE!6AguQNRm0y zbWR=MWP*(2m>L*!w#K_?tfkA+p!PA8F$Jmgtvn-4feH3hm=gn3z=)$0p30=wrJC=3 z`D`5Ns_WK9C#Q}gSkNxi4TuYJ+OVh6j2gjWA~{-y{qTZ#CXoM!f{Zz+(}zyw=wnxJ z2ao2-{1{SHpmB8tf5JHnP?HB%g~O=YjKEP!-6JSjr~~2&gB*3-P?ph(xo$zb6%kDB z(*dB}E{u-N+y2afqRM%n93K4-7hs62zVd;}vX`>Nf)3y>A!4`80c|yhv7p{?Wy*-g zPk6@CQUldr!-*%e#t+iCry&rznaDvbYC&VQYk4!5%Sd%4Ff+MPrWh+x<__iPfhj4J zxD4%D6o@}Z#|$7|i>>>m`%rS@P!U8dVPZPnbZ$faP2gk?tS)z4di|S6u3rfpXg)-M z%aIh}#JPpzsV2@vd4YI?$F_9UYjP<`o1v)s5jf|2*E}?J1N-rA9!$MK3dI3u;t#xe z12(K?H3j#FzQ?u#7h!M^nJMVls?(O82d3D#ZO^5>68A3Ph6;>EO0gVNXYpEuFgiUR zvJ6+P(|}ugszL$nMu#Bg&n;+<|BmCzOF|^|oP9g|m_i4@ko>D3UW1X6Dm!Q=A7a6f-iTe(2xR@`tS z?*meNV%QX^o6p9kDM{V`Ld6oeK9yQ$hxhqiIEq{N zuNwqJB+J#>D9O+iWrynLx{mSZ4ZlA6+PZ5JD;ZC8LNLK^+GzP^WBk6`A5WgWm?Z~f z1OZj$wf%?Xh#@ZIEcG7_3GbI!&w^%wm_P(ug2NuaPi4cmq3w2(?o*Xzr6i|FWmoyW}m-O z=Em!bqhsW*O%#Sps%r>Eo%+EHnftxZ_HAC{;{@ZYW2bx-@YL2_K48plF&HbM6?b#D z;IFhi8!>}}=x%Ie_AZgk3wk!>TRrbvYJ7ENU;4>K^Bez3Z0qU0Og~wDh5>T#GOZqrYXk@sPsP{7Lq+yY6N?d2(wmL*5`Lxs zSriLc;q*IH?$e!$+;TZErqZ;y%3LT(nwGmY4%h*5t!8ZsycyBCC#739-1`+h2+jA*^zFiDV3_)rGT>qFkaDh^V`UKMSRHZZS{V^Jt&b>&iq9S zV&3-wh0n$irXUrKNLr+PZxbx_#>GJcTs+&5l&paL_%1_u+Mk4hcJ6Da_RHR-5@Q+b2Y3^t^5V|Yw#L2RGfYOAYO8!h~7IN)|f3VJ$~TSOFg_C;?)rG zxR>({s^QDObBKV;4CtN%ffxUU7Bh%lbPy{sA3YAeCu1QO9r21fTH9j*{9C=8Fo>aG z2JP$7pi5;ExI=lwg-I9w)0HgB(tBk)+a=z3bBcp)ipz3jZi)WYm6@+}eq}^(Tiv)Y zm?dG|9+F%)+uo!vBc34q!unrY8i23)>l1Z4CE`wm!_tI1GDjWPIYK94Au#T>6Yh6@ z@*Z9qeG^r=v@p8I{)=kkKxHF+E+Abfzt`4KS2>|~;PTjz&}o$Thm3j z%4Q%wA)zOuY6BK|mkxsw0$U$`sjKY~20TbMVo z4-t9ld*?ifch`}uF!^}ES%4k3_q0PEjLL#B$G!L&SxCX!#`_n1v>x!hH8hl#s`v{L zE5n(fi-Z|$rLikK5ccqHF8Nf&jAuE8f&#CBHmv>)8$YU|GvgB!=Og`dw(W?>Fepq3db4P%cGT#cCjoq-<& z@QSyUr`j<=FlPP;g&prXR&j0D{#66Gv7a&CpG{ku({jAZJr{P;` zu8)1hqJ)5XZ9H&U0}^6AqcdZ{`|Wf*6%g`1hW8O&?NzH}UZee99ONnhjm@13=8NPy zosvqz%%6bq>#D%^Fp!xfVMheP#x%g*3ZZYs2{~{p5XPU7Fi~?*d=D`CYA=R=S|11F z)gS&Zrhw&3I}{O1j@K83 zuxI)}%AMGNpqfXXIl)IXL~w!2>Z1Qg)?>T(DhIOR1UAlR2L=Dmvby`E$JWR@K_utr zbZ^Sh7-P2tL5*-ekr>DgPkl9`Uz9m;qsk#|Uz#~A!qfcH{*w&dRl_H*|y4lJ14khd0&Q7yjzgsb`zo2JyC?UG1ylTP1Gg4Nm-3;McVKt6)`QW3jozWQ2RdjgDLL?F*0J z?fKL=^i)mx$Z031*+apR>Yp~=uojrO-R4(x+}!d%m~Z%pT~AEZf_*l}7ng-5omyI1 zm`r%3w5u`hrq7Enkwbz81(Sx)eUHU%Q!D%L``|c%ta5(;km=m;)y3iTh{+QbKhGb2 zZ~0}LxXmm3h_jPE2~jPX5|w33tmWMgMz5GY%N|im{Jfs_`;Z#}s1_dgynwFQ8TEEN zIj8n4U9l;OKltla;MvLK_YOB#`xz~d&XapJ9n3m@zkY0C=^Miwyh9G&v|FB1M`YL* zho^?)zVvCDXR116*rzY_~?|vDX7YL2m zsG-T!wOuOuGF6^1eZS_nYUo`ZtD|>~6fU(6-&IiD=^pdg`91cdssFK(`+Y$Tjocun zg62~-dW}9t2&n&;K$+EpVP;AH{KDai!99beU*|`*JSh&YIaPW7utP}sTt?2Hty^Cn z%=4|zHi}IWuB6vZvrR(>#y;hisnH{ve%&xUl~T6lLGf;Tv71@yP|CbXGe|^QDr8U{U%rdG8<3i?#;z*qfDJX!Q!a_)YbBsQQijlHD6? z6!Q+nJa+0#Kk)g@<`raJ(L&18o&i(k)cMiJF%7Ke=36u@d$#6XHQz`G4=U#Ei+M~= zr~bbZh8^?g$E2mD>ER30xER3?wW%+nYYr{3|6qWP)d!|#ng|1wv`Rj0Tu2~@#_i`T zB4)gnE4`H~IxE)@B85}S`Imgo6YKNIpOH-`hmvUumqZ;AQ+R2niCQ_BD~5?WHqD3r zl#?QaW=#>!$Sj(ac`-y2s|#ktlOnvyJ6^>FI#KQt74q`Q@m9@2ywE zHc}$xHfUxqzTdI$*MLK~?}bU8YNb}@M{oXEU!N-fTYFGU1b?K@C1UA#*#5p4y5@J) zGsz+*MsL#E7e3*Te&%|eWXZQ-MGYvzLA-n}KT?ca7gJTK ztgCpfXR`ko*0{01Yf5upQTqE&tJLSy=EgBslv$0F{9jGMn~2WhCGPu`58(rQdY2cL zB%T5OkJgA+$M_Y)zB0k@lL?;k{LGoK`uVAhsp0&*^l;B>03C5VG`N?ez!>Uu^C*dUXs-&Q+U#w5-s$W1n-KtVh_pCFKfBtT|`J zzMNvr_vE|#t%{$6R%VuI%Q?BB1Ba?o>WG_CYNpbS1t5?Ad~5<4&n{g_UrOMwa1oaA zziX{kdY6^=h_%w+L!JNqlunTm=aes1F7IFNysZmsri}NtYDU=PUb4geBOe63(vw4j zUgtxhN_*%l_O^mmYppjuq|X&K{uX7~T@7!cw*B|RYL9E3&ivSGxXWAM+;w%)Y2%41 z>OvULWo8jiJ$e2WMXBKdm>Tb?ES`HkZk+r6c98H=wTOpuq;`ktx?=g39?xTSb!NE?ADdJ6O(7yXjKAfxi z8Ex!1{>4Q#LVIrN5a{}T$HZ~^XS%Y?;`s^Dkb^y{(aLiob)u2glwZx78ow>BKTvoP zKsbS_4n{A>S?39&zvH8z#izR4gLeFuzq-0|h)RHJl|Qe2)BOED64`T9fLJ6fF_7*s z$Y=ljr9^yqZL076l|A$|#0R6Vmft@Jv0Ytcihesf()A$ze0n2GPfXeOP~yA-c31!B zoxESmT9ZSwi!R{Sm%Wh^@r}anADq&FS3U$a_+gO^H{E zZXm0^M7sl+`{Vzc3&Qq(<~Azzy+L2`&p|nNqr;$Br+JKoOso)GU5IB0@$``Y=QqSk zxwT(?4IZ>phgUIv()d6$MfXnz7!U-FgcBt|oD+EOe=AKK)>NNt{C(!5g!L2tHr~U^{$(LpMI|(T+s# zmes1}W=B$i!nE-lB^VFeDm1ZNplcEC^8+id^pC0b3Ai3PL(xYu?a%W={`HdrUGIeOlfPz~l*5h0c`cC+j$Yky+Q6P+774hs zdv630`(~&L{+wQU{Z$FfGoxa)AByM^?_A)K<%wEZzqP#Z$H372Fi@xudyhHT{_|>T zvByaaS~Om|7_z{SV5fgxhcD-VK;0w2YyJpO*eZjG_l0y$);BjLfLDWcM8>SS0N9*! zw!)nHb+69Yss0CF3!LZ)0`>$*9$)6S8?XPnh88Hk5nWGkB^&jCOZBaT!XUaX8YVZSrhKec)EE2}MpPJke15L-0%E(Oj|tI$;iazJ!TS%a5D{Dz=-lLpuiz`YD%dB1PA!tcc^Qyv7)?V zTbG@wCOW8o06p?ln@2Gv&1)uk8-%=-%B@>~2${^Z{gpfYJ~2Hv6k99c4T4 zD;s7>oF_!u*KqS9$(FxEmMu)+o+&@bhJXCh#meu5fT9@)%sS^{HM(fK0YSE+O$1$5 zW%}Q@m(--VvbvL4`wOr* z{&3=*Ych>xf9gDeSV=hn-JMy`Omu@R)T-V`Fm+LDC*oN&(n+iqJ0Uddf+wC$!h`)v z#fXTP#B~Yu2PcJDBZID(^-=ugGiS<(;_)UJ7eS@qz>3AbUnDFJveE-qWTfzZ1Bo?u z*;zTLL&t#UNwb&Wb~uTH)`%IP@;FJDyRC)Bs@t;;EzlVtjf|Z%N1(vM?+8l?TA;3g-0n7|!e&HiX!A9xm7w$_NEybt8EGW*Kh# z!RV$DY{g4(cAWMn=q zxdt;>c7&XdAkM>xlC=)--gE@cc!AxZvW+bN+CJNSEF6z^ke?{AvcuiJs3L{$b~*@h zyKo3kQ)LcX?$#6F7SMq1Nun|!;I}W%L^6*4AaZVb>oEh97JBgzRXz8vIb!>q#2OwX zkBB%DB}FxKq-z8L4fi(3;A(J2JJb{gh~zz`tHT5M=c%CjIw*yh)`H^~Fu#Mj0&VSH zRQAkRb?j|8NA)e--ws;I-R>+=tgNx^8&<^xB2-&*z|&7AtM{zg`~u4&gRG*Wm#?Ni z)`McP`%|%?$pW%+hk7{Q6rr3vSVbHZaYdl`NMAvs++!1JDu%gyPyzW{@QW)1XcbbX zfFU3&!R!)N)i6%==4poxTU zk|U^Ys$p32RGK>G)a&+dCsn;NG=fY)4kiPpCDqm{filDg{d>McsV7BX`&WK9zDWV=D4pJ-mup^gVuG|}T*6wYvx#Kr1s|J$=F3hz&A#4B; z$o|KY>yikkAE|o?@&)jRaZunJOpV6v>!25Z5a2wM(*0|G3+ir7Wf!bLgWiQU0b?Uk z=@%nfgZRmx(?E;HXv$v%`KH)Y$TnadSo|!fd22Ujx{fzxuZ*ad1^VMd{7J0l@IiF? zVOzMet2_3?wdDlX?0!WAJ$w-8T&y~8q>g$aD%j`wgCVviXogIz$3@_Il_{zhFM;r6BS;SbRzBkir#g7#_y04evcq~ z0?F8fUvm+l-75CtGlC#+MQjuvI4N)tUIEOomxe<<4pD+LP~dj_p}P>AeOM3sLDVcsb7f~h7oy8#Pq}Ui+nbd( z@#$ygU7}86H+F;0wva+Iw572)V3ka=jj0QQnvUUH6+hEB>TvM~66Y%7EW&$6W39J( zvWzWz3*_7@SVe8zcKBU7zm`GJtoJ3ge217SkYzGU04-aeOvzS`S;g^cFZ5GiZGK$^ z;zJ6_tkh}sH7u`pVpxqf$f8O59}%Lc--OtI^a1h7f>=e|QHF6BGgU`-hhK(sr_FM|XB=?bvhhWRabr9QP8fjUqrqpGR`4ibdk;Cff21O}^<*yG2#Mmgj=Qu&^ zE$8bH8|;OG)4;NQCpx_&SaTJpa&;0z2||<Zv#93Nf<$?fFE5teHiGfJjFsPuC>jp6x`X!AowXPWmJmH(Zj4^ zV$akAM_$UYr|x-qtr38j^=lE{9p8#;#afplp~4)=0x zKKq=KeZD5!6k|V3G|@t$1P$TbCy4*M{A3tpw9$%Dpc+_Nbx#;^Ru^RW5Gi+K{95$T4z0l5 zIKcbAog~0A7r+z|U}^S2>H&A8xkC)lBZX=;{)sx1;ljDMunIxz7VRb~1PMC9$_V9d zgxEWvE6N=jGsg2~n}EwmY>g?oYeBBB;V=EWnIH2isX(jA$On>2`bX|?uuO8nhii49 z*wu>0WqD*`@_FJQdB-15G@7h7Vt(Vw(i7F9EXN?cc^`g}qqK(G!>xjJ7Nb~L{r5lc zygTQaYtR=I&_F5v%qO<%XNTlazv(LkKOQ1>6@7H~+IE6?uo+kh^ICd-K~g@~z_3#! zq}5Ml(rLPUi=W5vFLc&)hdSoB**5+>VwoyWZ)-PRz75Gk>YZE&_i)19j6#rbXhDuG2Qj5pMGoNfv0kHPbLt0NmG&;F#YFSt#pBJeAZnPdqCrji?>kLq{$C zU3&69O`3bsRGN`YX=}&*fX(w7h~rJ?hBnGzovsnW7kK3#>#YZ!p=2=v-A(e`g$|GI za1Sih;Yu@JR33Iat&fdn#t|4Tsxf}O2jOKnK&-d?-EXg?AAkub-|Chgc9nXiI8^yF$NI3Uh+uBpK~AUhqB31*MZOoG+y7 zjYHW&RUpUQv53mq59O%B69Q&Lf%@G(8Fti-e1qX1VFZQ_3r}U z3tZfi%p__q)-ukx!&2^an0Ww1oiHIRkt$Ge{&NeOMCM9lzpK5BwG%Kzw+ywdqfmR* z{Xz)(ENF;Lgmj`VWNs^2Hw$T3oP%nWln~|!T_=Ig@>KAHYGdF{kfrLb=_E+^q_%L| znAtly@)^wLboPq}+jgAHgZ$s+KFc8&Tlc;LC+TQEnQ2Kxig-&+C{aeYSfq~`MC`V!*T>9m zr}i<&y6M>rMV+vAbM^i z9%e&OoeJU=A5B(QCTPt?IyO#3jn95lZ14p3G*CyFKsVQ+MH}@OjcI7WHBZlw)v6Im zLakdk!(vIh@T=!)?gjoaL z|7zttqngUrKE6*5NkmGNuCx&m6%isuP!O(tkTxSNK!9TbMrsfQDZ)`_1`(A}j06!F zRA7`AKsv}#N2N1TLysabv`7(<>fLzn{qnANy=%Q|-F(W2_ z7yObM)zJ8^MG`5H=?^k|35pY`#B9|@TydNE?1TcouooDn%xoR?D2i%XxICN4AOvb7 zK823jfN`K}0~W_3p8wq<0&)ffiE@NR>;jk%e)kfGrl3#z+2Wd(ljRGxMvz*5>-ykp z$mfJ)R^-Y@+pyGVyzTu(z*-uIicfjd3&6}kc~v?cl=`bkhlxKVWGz7WR`J5(hAq&I zYiT?bx2C^LxOo6?3lw|4jkvU<>Cp`hA&wl;3Whe~r4(=XmO01Xx1Hbg(>ooW4X5S> z(*slD3G&>);of$C!kbJaEdt)Ha=$A`Hy9Q&uO+XowO#PFid?E0`D``;m)!D!p%{uq z&WUm;phyHZ2@?{Qoj!=2Z0hWQ9HFuRmWtuxQSvcyBO%OC(AT*QyN}YXfQIW--%s56 zj*M8Bsg91h3nPD7|8)B*X)cgWW#!O*@>mCS?n%>wp9+B5V#i*mHeQ1jsmGW!HZu3~ z7GfgZrapcUaf(bslo}bC>Y(U@;Ac0`vVm|fA~9p7raB9k@tTKtudoyFzV$}|D*jT8 zKzGW}Ey(pdV?ob=0hu+g{!FJb_pk(FeP6UBmbnKXR5n#|uL%P7kMP8T_ko0+l{v*b z_6IWL=6qxC3lT~4m!${G&71*VVariyn& zd7WzWN3Rrk%klj^Gf`jZ>eelrpwrb`nFHjj5)h|H_TNV%<{DS?@N;CM@y^waW;Tsm zGrK9ZqHV(tka|`{1mi&>^tMNl#|R^Fy3gM4l(D>YG1PayWiu|Es7Qy7P5~T2y~SC> z_A=hY&`lx>ehuiCM&s0b3O7U;7aUrAR@{CaohfPx<)h|+@kucl5@^$wY99UP_u=LE zlX9jyb~>jTt_^>ER#aZOJdW2uzD%UIXwH^b;0cO!bolM(ok-p?qNJYCs})Cx=V=8< zd)NjX?hDl54gE^QL$$DriEYr1ia$%oJ}eyfGqcJ@A$96x0-_%4xQt)!qSV87d>*Ko zTj+^3S2~(#%fCmMg~-g0M8Sn?vRqw#5-(|#>}~75+d%`{CxSQh`OkHSxKO#{TF+In zT4&3o9_&sJ@}`w>PW_$H7!EiaL{2YB}T0dmiCxoRn+ru~S4>Qy``aS)1C zuF&^i0#4n3ARQ@MpV6Te+;joY9vg#;X_hJFGZmTnN?`I=AZh0#%3VDrS>UQcKsFU9 zPZFyasFN@(uEbA-jhecnd<%a->H#`!WXF#E<{p>IPJxH z+_p%Z<3$)L(sTX(%vwZdb`TGCL*hF3aMrsX8s`S02C*l@i#rKH{#u-Y_PP>F1RorL zRd$HCx%tB+|4_{xWzE_ zeaMQBXDO&l-M#w6qk2z;I$vFmK9Aa+*(St=&TcTCDR}WLL7B=M{hmsHmo@%a|7*0~Pf*pFg4VND3%`=75J`3#{^M^!1v*ar-QIt z7CQ|WKKnp$#0D3mMV2Tx|1>ZYm4JY4Va8cvwDgo5y$O6-zG5WGa(2{PkINH_M1nAj zjdHte9wJV~3)~ejw<6WFrsc#^7|PcQO)JASur4`u3f*6ZI zL^?9#?p9Cba33LaD{_4e16uOLYJc3&CA=+52HTS?PYhp5kfpmyaTSUph;X=tz-yrK z6qdb-HX%ZAyQlRK9F}y6MMsV@S4qbS8jf&hA;gzZ_gvSFkG-9vPm+et;DgTAVwh&; z1Qp z))NX`b`>j2{?y4{w-udx2iGcvmt8<(Ekl-`!A2RunK+z_@A@SVhh~{RJ`r*o4}oie z3}RM5*ZCD`3B9-Kayc87Df}eF8dm>Y7gx)ty@WsggC!*3@?JS^_>yh>U06|Ses{Ua z3;_J=DC?E3Bt-P>NAOHdtnm%fvV{oDz%O)tLbrn<4H~9|a&Do%8pptpUA!fz@0PEG zR`O;ups{L7GB?DdA@aRO@S4Iq_2@&(R7oU2o~{SoQpDUXq^q_1IFvoUfSHX%qgpyD z@b$Zlmr}WlI%K*JqGs1%tBvj9wNl=>z(5;Leo$2I{gL@YPQd{}$dEM%NEU@Sd6BHO zmnS-Oe&-vckLiVfEbOG-PHk)<>y_9gQ>l z{+Z%e-JT*`X6eE%DXwpbC}yo|R@Q&w+>VcEmVSl^!`nihwPMQFCBmiX%4mJ8h=J;1 z+*FBR8)t|CW}VI9gSj;#yf}z4?{P8tdhHj$Lpd-@NBv2-0xJ!vn?<;!dQ(kew8!QK z{LXHk8;iYFu*MluQP;y-ZxQ;I7!;a{x?*%ivbhr;LgSZZ!<`9jywvfrGsIXk4{7l| z0T*)TLvN9+n0A?`uMx8=M+6L@R1GlS3OB%XwY^UEetO4it~pRboMV>fj|=z9VJpAY zBMjGTzfT_vMtI>hkYx_{C8S4gJe|##!IG)m1Ucc7C+dM`RKSce(X+rN%X3}09+e3n z$mUQ;fo;=@wkRhv_IxM)T4~51cQf3^_ayNJe>iF#rz%wJaRAn`+$EuWVd5;HM+?EZ zVccTovrJ=RRzjR#3x zan4hH!5a`z8qs!D8WcBE4RpbS`#%!wWnlw<9S8V#9om0V8JoQ+^z(?5%`hd)u0#s^ z(X>%B;Xx&7PV0p|PtNf#WeaSYP~$soU^*4pN{2hHY>yTK}InY zzYq$J=iX74Wff}khn^2?A#7ocS9S(4(mQ}(utOg_`x+};lB~Ol9*?&T$q^MO=X!be zE>kNs$g!KsfiA*&e7lDTZkPtWS~of9EM$IS$qd%V^lgOYsAzA7EWAgV(E)Byp3ujv z0M8)}m@@*dj8U_wezJ5L&>&=1=!olVSD^G^BXOhI&fPx8})rYTwIfw2?*x%)^VhM zmy#&&sh%jy@Vp9md)5y>UZBb^(KCk`FyA&hX07V=_C2~q@h*BZ-mr{^s6z{4B1^Pn z#z_T9SZ6Rt{8T&JFoG&1n&RD67hb)7M6(8&`wrq3ivD5+uU{PP@tMC5 zYw{sblGvEr5KXmuTXimSGe14qm{|0h)+ck8faIGnDA0Te8pV_J;JNysf>o^G%j_&= zGEAj(P0Lul=_Jf**-XP;E`h{LfB8(0I!Ln|3qj9KILZ$OB$K%Y*KQ$0PAHUNCbBJ& zikhr-26`xto@hc0(w>iu6$ThP%Y3cc1J&22$6GB^+iJt{hDTe7#yK>e0rIMqyB-B&7er5XcY3H_DF*!1d!RN#dq9QO`if zD?8H!MJ_@Y4-5F+-x$gpPY@z_+>8|$5~5Y zN=2?_a_XQ0#xEMS_FsjU2wnj&?n>y`=1|u#0R4Lwh`?D2KLf$r6Fc!)5-?JNp^CM@ zs7wktK@8UoqRX|&MKkzx8x$2#R(I=bowC zr@&WfjB!5^f>*{Ny!8;VZ;;^UxRVv6L8CN$21;sN zjv?^HaIhOOnIaRVYbY?#@2uMa;8_^ranG9#z2>ynLx87q1HY{x{zm{$S(Se|0aUOR zXsjL|P=4aM_)g_Xm-GK1-irMXiunIQ{=mtW_VIra#d!#S8d5`!+7F`}paP3RN;rM| z(Ydr3%|6Y_o8mjmE)KDWw@NC4z6gKN=%Rp2PWOe4u{F!S{K;;z^{aW^;S<&fCg=V) zbXa(2NPN6e*U;1Aj~^fEc0+=8`J4idosQQkGu{|b3leZZG3p8Yp- zF4uLcBc)X_`G@7@8%v+1zZM52{LxTyJKM^0S#l*SxokQQ^0+~`8_c6I11Nv2CuBUB;ZpwY7rV zI~DW)0pZ(Cgr|I)40Wi#Z6@1!f6>Fm&OMc1QZ$=lw#!t7ET^r#55{>WyH2i5w>`E{ zHg|M8I$wSFUu^IxB7BimT4h{iY`S}OIxuwFZM5r?X3EURsfU^&$+F!tm!aDnqVIJQZ^)DP^$CQ#*39oHSe|*%zf_}B+4-M zu^i2_gTMUr{kMNk-6@;?EelqWYE{&16?liD1wZw73 z`O@nP)}PRs(_v>Fqcobo0pSxP%bPz6PWDF0D(Eq@nqNrw>c%{lj(nQqs489TnDWE& zF{bv$d5vwk;i^ymO$Q%|RdsKfKQ#9K4r}R2*_2&hXyWnDhR7YiAOOZASwcjAdG?K5QIgL9L9)>ii+X_0>=a@ zX$65L&ruOjf`}kUT2VlP1j+dh=X}4v=e_rN@B807Hv7!>c2!UJR99D5eafo`59}2Y zk`V#`h?wo$br=95hW`NpB3{$RUz7&`8XvScvK#*jfIa{$psx?~@y}a;mA7{7T1`z& zymq;6`C{p_Zt1h~^l#y%)pGgrvb4Nu>E&hVg@2}5-mE;La)pI+(Z`EEiv?J$TdZ4o ztXr+DJiIL7t-bK;W}4T{O0P7${J4qJd*Q|F=1t3+X_YrC@$%zl<M!+2O=Xt$BE(r5`w_2aSG!p9kG$i`KNZStwfqNptS;&6#rZ|<_J>pl0`Hd8$oz`A)0Wxpl&tk~ zDm&F)#q`-zg&M8WB|2~I;>tBU#K8{n^_^2HmUb@%R;(X6a$C)|k-T@sshk345fnK8 zgo>^_^#&rO7B-ODp(6+O+!Fc!v&y))ynL!mJfe2ys&;uisut@V;os43GApt-@*)Y{ zj?bGcH~#!+_RiAM6+t7ae*5+TPQ={cBQn$K??8LwPNTG6i=*Gli+-RT&>bo_Yr%$^fDqU3%$`-zN8WkV>zFR~TIBAxOKA%ddv;L?Aw@dQ9 zJMLFFo+vh)z85cqrIoiTEgiJ_Bl2SRL3ChBDq_5IvDXx+ukR2II$rDY(Da?}Ml5@( z?42rm)9BgCnJdiXsO7QziFPjZw^~8o-9Y^(#&Dt5yZg=0TcdJ9BSBJpDc%#jHXJq$ zi#+~^!F=nKhI;8#Lft@WAW&Co7Y*`(*jD^mzG;0gCw+IR`OEYj5wp}!KKpm1bCZbv zDMQdk`u>#rZ#Bb7h_+mHeT}RrSXX(OhzVqW1B~;YdjES>wcvE!Qs}$)x2OLyoumWf zgCCXVXh3n}&sz&*RBV-B_|ddYGTm8 zPZ(iF3XNE3&~W;S4y&6CRbL!E%)}qTykB*stT8o14a-k0TlnEOThMVM+zKCFIv82p z_D$^-k(qcgeEw;1kuv{9{N!}F1U0bWmtUJgT#^MkoMok0(v_8EHF=x;TVnB*XvUsv z(MF}6IO&~%)y$rnd*OnaF)JBQw?2)mAU@qIuK|k7evm^WjAZ%jzt^FgKx8))HUHCe zKbZal@*nD~`grByDZ+fue#Z)*05|8_m6}cuYVr?Z7ptwk^T5bKS)o67?L1RL=)4lc zOnxvJaN#?!n@pU1BMTjQ2vBf0g*Wz9`eDDzr-lv}-}vEZ6-Sv4C8E88KrHtmx@Y}* zIw$Y^O;_-T}R>8=AMT;n$Bx-yeGC{<&6|Q%mJQ_R;-Z zBGxVpy}UCnZvl0Z|Lx@y+Rc(7wsi&TR4_+Q&`5H=^6#(zcMNJ`ye=g)ab+11+bux( z&%orU+-GOPFprxO|K)KTG|Z!0vj0ZE$ZiCNpDz<+Zm$4Icdw%Kj-3Bed9EJ81Unx= zyDeCh(En7f8u{p;V+|_Kzb8!_ZaqO?xBWl#w?qIdU6HGuT>m}o0hM@DB2)_-D{hc? zcdb25!rD&}t0^Esy5`dLj`tz3Xf=^7VZz@|BC>-)^YfsySnt^z?tdHsfzL!h+ktPW z%e@Z-!wgNe{6%>sUS+g3zwYCW~Y{af{f>)uDM4Yc1=4u%Otpz!6 zf#gKf)5~BWirb2Iw~Hco1UQgeBmfT7{YK-c-1)nSwC3^4gz7FBo>Gz5{eO<34Ggc4~DV2`=8(k zB{>03Ac(n+mLD56PdaHG3{!^$4I%r6dK?*T1lfOxM04fJT+Iewy$+k1o?& z8vB{kA-E`NPv$0_d%oqEx7Hg?z-_5)fvshp9>mP_GElr1(-LJ=iA?v5&(9B^9=p~W z#~;3RgOKT$aT?q5X1ffyT%lvbxTKtQ5vx5 z0(QPcmM>HYA`cDRsz$rt$Wk4z=0ZM#T^xr^`D$2~3Gi~+hKLJuUIBB-81O}mMe;nI zrG<3~S%D-$uELk+Xw@42W64lPWRuG zZ+!Fs?fxu^yq^@A9{fTGEuH$3H9cT;K+z#*oRGO`P=MRk!cvoA)CllqMs6ENtR8WD zA`Ng2aKf0zJ!+wbC`ZKY{iR+R8!!7D?WPY0zzo*Rv%InPJW4+Q9Ie^owuKdh^;AltA-}yJgn)#Yq%gaoZW2C5O(wb}UQLP- zMX?ZR`$c4}LmNMD#M~Y9&@>-xe~>!%!e|TD!_cQNPKLnhFxuu>YJxmzg3(^`K zcip&p96QsJaa)|gA#g2ZM;30(rR?5_Jp6+Y+pZ(P*+OZSGjl+zR$-Pd2^2>+?00C+ ztyRkkKjkSwQWu~3#n~TD6)S}}eX4iw;|l~Nq=W#8MnXw;$9j-g)_3NxGSoROj>d^_ zv`~A%4{orgaRd9dK<*3t@N_I63lN>2xL34Py$+<++GTk?M%@e%4;@8h;BB4==jd5Y zV0-#KfwLX8Z$VoiC@Tma|1c?RPKfp04ZIlIM|Q%5ZNQO&q?nvcS-cJ;?aT;V;&jSf z(kB9%ApOJ99t>Pk6k&&*I%shnFYv55;nO}*pb{hdX0Q&)tc zCnE3bW~BZ-8Rke~r-a$7MR{Mf`K{Jch_-+-K_?~{IA}F~dtqZ`i+PRyiR~h0Qb5dW zm?I4u2+%HsXc#{OH8n3hSl{_AnaYp_>64?oX-5>PfTfNU9>*+QPl0MqX?kI}V_u9E zRZX30NvsCUx5CwOXqX<0T)h)Icsm4r{gd8ofSgr-J~ZlwO@vqkL5~Dr_yAy(BPNJB zvI4Q|TY;SD)Iw`spH4n=_yE460_>DAmeNwFt&WV1&K$z-QMrcZ>Gw@jKuS2~C&Ztf z5kw;;h)k~>d`bP03jJ9pk}Ad}-bx|cUs-tQh;j_Jycfk^P#i&R?wKN2FKmNOZusE| zqNIbz3a;VmdXVxtJ$Z551Vo>bV@N~KvOl;M=&%0x1o!h~&Q=Y6kD~-g1*|xbL&A17 zt`_4dNWCH;MgoOsx4;M?&v_dz)>Dw@R@6rASc&ZX67SC~&c>=1ya<&+4!xu^q^XW)p@TyXox462e%J}*VsUXh z?M}Y!1l)9o@=e>XRSt+dRV#7wH{iPv^;R*~jkUN0$N@WhloFsas zkcqM7C!xfn@Rt|tsaD6Z?8W1l8i7Fr)!MjC2Mj^;7G4A?m&T)+8Hwp7H z9ULVYHL&?r@)2Xq{hlm##^EL*@0}3Pg?!t54agg9p;%m_Ru`hJaTMlh-!z>>)_y}| z&h0N^^9+d!5`>A8#I>S~*6qlC5U7J$Zuy2EK3n5NIWs_a8v(SsP@!)0Tj&+vayof7 zM+wU|2?k#@FA|%2SIna|zgeIfZ**VG9(_)1HdF%aJ(y(uYB*3>ggR_#UWF8l>ma!Y z&Y_KYstR0brN{V2O(iH-e|4{JlP0urHWS%7*H05z*MmR~8=K{q{(-UNVRQ1In3@XG zPf%`RZ{6@6d`eC2BK?%;Q)h!r(Vfl@kuHufp9jZCt+PAlF zCn|gGF)Em2oBAhqCbR;koZeOA|CB|{5aRXi%1jP{dTXtruKb_veSWsa*e=Ju_j_PM zt`ylb>!u=CTHM`?4E%Kj5&*N3Nn@N2zCeavLx3v;TnUCtskcu+UD*d`zlW|llEp8? zr%L5i8Y*%kZLJA1CBS{|P{Zto#aM;l7fL$q9T;kvWYtd^f(Lh`umi4VQ|tru6T1ms zwTvE8I?GXkp$geKl3fz7%z=NQp(2v|s6FVcL3wJpJodZ9f`S>YV&9{23%%ZbKA|7a zxg(FYfxNmKMo-uY@{^2x3Cb>j^&O2PGNX@B*_onMw{}8_fcUei4ff^)fBFLn zMx*(i7@KlXEW}UQ2J&KP?4s*^JGbsncY3*>PURdnK@P7$a%T2vAs)F=6i+)%QQlu! zPutXyxq@S-LtOQ?z~%+zG$NM$I&%l2|FSSbK&JXimCIYK|8~i4p?MeSagEmI{ApU!JOa31FH7 zejEDBCI0IbgZ27oJ3M*%hyR-gCQW|*S}AnybAaZPcqLTVEyqa=Dxz6q;Z5`xs--o% zNBOEsaozq%!RMC@6Ea?%#D*-Y&@^q3vmNR-&N;FYCvms_k;D#Qlwah2-Tsx}J z_QOutq*@F3kV``1R$>>gWot z7wrxhaV50PXdO003hl9Klfo~gwuWi{8^0XDVBJP!M*hkSv=-qA6H*kgK@3g2(f)oG z4V%ziT}NqdSPMkS7TbWW;_5L%X2c8yr4!RrEU9gAWYiV+R)}JeWXBm=)p;P7kPB2T zO3{4U1}V7&^RJjn?Ef+2l7+@xx%+L#`z99yA{yAdjfIImgT2{UYeJ9~4Kv z(s<$xXJF~lLU&)ihEsurw^d!FJ_&71R=`f&o!$)i;aJL-Q<1l72+Cn-pA>&Hc?3V) zSi62rPz943RV6M^*?%eFVSOe*<+0iiu|w~#qZ3gxk5Dr)q}CON2Fy%HjhRFn=cptd z$|5#7gD!}+9rUf5zFOmKUNwwy<+d_cf~;Etsszu7cr{jqckCjg}F|0#?n?7a+P%3(E$#zOtEY*jB|{@j3OUqQgbg;D60avE?K zxHQhQme;q`ZtubpW0EwiOcu_UmHJcgV|T)h%<$XURPKrfLN%RQU4#}WvTa;~$l>-= zHQ8tl?MpUV{`fm;;F;h+CH>o@+5nvp7mhulbNk*)VZT!s=Fv3V=$YMKmor3;X&Nq$FlFxj*8o{A9T(3dDR!RBBGtT6th zAZ9GpZhsQCe)u5L{JifO+NxsI>3DS^cWIRXb8^2N-+dK#=y5KzYx+I20%`Z$CeI#w zH?M}aw+xznUP64cc9*TkT|CW68Di}A77vQL zfn=kC1b6t8Lr2_41&!Lv(!jHDErlUQs!sDnRcjNzH4yQU?)KPEZL_ybbtsGhfsH@H z6sIT0UusK$4`i+a#;SnmMA;j3!4T>x{Ua<^&Z7}$O`>H)IRQGKucNB(HHn!tz{MR( zkZC#m&5fASv7KX`Ycs+_4OEP}&D46ydK9(;TMJQuJObhzeWff8N{H}g-es@HywRC4g@!;ybPa>*gkd#R4LAH%3f3jEPil)QB%j39>>9&S{?FPYrIA2)|;2Ocj z9!#o`7{?UY^1wgjlr47m62u*N$~ye_M?MnTp6^?z`K%;;n1p0rE-2~q#PCDoertf_ z@68isMw4wy+wD%G8JV~^~7PBzwbUn2>Ps#g;CX_T+`Ap+PL$8I41|aS98r`Gi zsh~%aukS#yki|leVO>t&&@iKqT6;db7&-RcXxqG@5TKEth@2;|M;?0-tx0SOd*6Hr zHi+^s9}M_~8>_+AfiUXW4c4`3U2%qw@*Hr;AA`2Ee`Z4&*#;SPV7l1*sb8PqaWw4) zOc^@*gMzE@0r|5h6z`@$<;fr6*m?zqTT>B@X?*v&>9$kyn3@73nyj`t7mm%U!XqT4 zfxG`$W7~CVEuHM>6ZF%;m>9@a@_i_)=6U?6?W~NGEFQ$K^ZfeL7*+P1ltoCh^nAt?N2t}{Ptsrxrt>lwElP@=+ zS2x)<=1xFcofw#N2xGl^UT?SGmSnL#pHjGdKHqW{_i+UYAVaX)99TR8y3u}czR zQ?*{EiVJXZiqQ!#9e~^6zhP$ape%P9S=YJ&8BP5Sm1(f&7`8ei1vJS*bPh^rN5Sd< z5nk6*VrH$m?w$CS_-~m`tnaX47uFW6X)K9*zv`t)YMFP1K($DrHq@=j-=@iTKaIWE zt$z$V%BfE;ZC72*&Akfs7~Mo2dm+X!$V~4Urg3d?i>l}DVKF3LjN9DXtAv}u(aM@g zEge>W2fmM5GY820J~ZKJ52ZQ=`WDy%LrQb;Ul}wGLxI#>wG}qBp&3uF6}*^$QNPI8 zNmTG;kG~|=e|r^nN`N6xPFFd-nyszKe?h_~9D}bx7%GJjsw2VJDlAk9d*t(h=p{V*6xxk}9zk@1+=)kQ2uPABGxjI$ z8%iN$P6}}n%*h3d!TAuoTQ7i}$*-Vt(jh19mk749MGz^~L$)8ZBUXn1-Mu`Eh-MvG zWbD;y$h!v9L%2lF92m$s4Lqsn8xR%y-lt3p2~$})QUHDtY=}2v{8PzATK|I6nokii_#c zPLkhq!x9WpJ$t?p(JilyP?yyk2t7_19Xti?q*R%8po(_mnjOyQ>slG@@U6{Xm_c_PiH$Hu3D-D%8_%pJP=C-xSX`n~H{ zaP>7ezKB(s|4odIfXoe4q&9FAP4mMV-O&lV!&fim!Lb)~#tmRWg#$HzL0=*-p2X3_ zkM>l46^ie5$I9K&0y@{}Pakw5`f#P`s2ZwT;IkTu7v_u|#2mv~E{kxseFa{H$O^`B zIW)1F#4HuaqVR->ZT9$bp%Xa{Xj&**K)`JlGiA&@s%atwSZg4eFv^$>KX_Ydq1!dsXL2u#k~8mJT5+PH87&U#``;hKDCS7RJkT(6dP z1#okbfECZSFQ79?aEyt~zB#`GO$)?|$($a5OJY>UCy=QJ(gp<>R>U^<_qSnkw1*sK z8Fds5qOlitm;;?nBHVop$%U_s=9uwZOmuv`QZh(m-B+ z{dJX`uGp~#F;+noXq*9^OQ4A+2CHemiwu!cGS;5Qrm$)K`VHqsL5F}Qg#)7q*NY!R(YYWzZ17-q<_YUAB$~8ni zi7B(KTi1bv#-AXQGy76HUXa}gQf!9z;5Q>ZPU|-iz2t8cYLDgu-+=TPS-qy%6VUEO zjp0gmqB`GwFM1(w{0u~_T{iL$7+lWTjakZ`2V<|H?^g>UqMErZmT04@0z6+hiZi@x zPbIt?6^CD(jcnBU*DCKUp8gE-l4IejR?|Q48P%=F{BZRau;&V?C#96t%Pojav7`-{@ZY8eiWRW1ri~$|NO{nvD(vWP*~jsoagq zwRZ!e`!@3@0}n|kM7Y%uv4zuptA?Rme9?yr=Lbtc<8evskl8-^{6u44fwuKpTpLq7 z-9>;QH_i~9@JoHZmxOnkX{G3(StG(Ba;t9K3T`!gT6#p{vi>d&Rlq$HE6foiPG%13 z@xPf0gCVPs;x=d6hk*N&>a7pZ#y3FCH)6|VSchZWFJ~Dfs9g>89+2VJ1_t>~yfXL_ ziq^a_AbRF&332-c8#T@jsSg}`WHoO5DP-+;G_lVn4h<@&gKFtc4V)s;KU1a-G(fterUi$8u$AOf9p#>pW&{{0uGd!omtMN$hoF#4ZcJW;9RN7`|;|VfPUY?_2Docm+=K&u3y#tdc#W8vakuo8kR6u<>Wu% z?K>93#IgeMVLRrq>F*Wznh|MsGav8z;T>gQeGP_&=m%i!F23IyEip5}!~!@FJ=aZa z^Kz+khiqs4@7*q+?=ja6^DZlkv(vT+($whOv!~NSCBuBR!%wHxu0X;<#1TzdkQO>N zlGr(EINsIOe&t$bh?wYR~`z_veW<)4?P@t}&;N zepsUCO7`%BO*7x;==6w5#Q_IxzR5hlt1BcTOJSy^eCA=q(rUY+kDZ3|i%U~hVhI%? zM9il}mG8xZ76$S;C5{COI{gzM%O9ny17iMoq+{)J|UAhP7OQsSEsArB~ zLE(Q1p^v);Vwo5*!!uQE#;>S5_>?ZnXoxJ z%%K9?3)*DCWC{~|Y}rgi@W-tzpp6%~{;vHZ2ALGJB#y$VBkTPoSYj>ny|*8>$f zOQR1%WvgK5=#bs~>9B5_Z|In#@m2*e+4Cpr8fIj9rBFOAT;BiO3u@_7^u)>CM5m2l zPy}SjO&$F?n{`%lb{_jOan>FPV@53sNSKET>M8)Nst$nhr-My8;tr_?b`GCD;ENq0;Mu>pX09iL9Q6esDp^P;`r4M4^2PUxR{Du+nn=(7K`#1UD8t4EHStLrb{Nie!_fZP9p zLBqjW%cwG()&B*$Mj30*5xcBF`3JdHqGv)IswzfrZ$B@(X^V9It22`{EQF^2%RB=O6{Dhp@EKIA5iNQx&dos z7m^n*jcLQ8py85<{oaMsZ!gu^497-gy&CxwlS2az%Aq-Q=loxXAisuA8-mwwF^8oq2bQ!lKQ839b@}&wyf&m> ze%;FH>FS8FtDMpAZ=XhO&O9|BajjG9%?6)u=c2A;cs3YqduST_^Y$w>67D@3rT0cC zgh^WeG(7r4GJk)ht)KebKSAbxQLByN;qSs`uPu2LS$uYHt87xyUc4G^Yjtdz`Dt^j zfAo`pN^zq*oJ*kdie-T8#HoqY4|^5+Q!T9ZUp>;bJ8`qZ^UD{BAO-owkJH{EX#$GrPzn$+bT4lN9FoaU|W_qIqx zbs8=nUZ(ri=Uqn)HHpx6Yox~5wxood+8?R*aozJxQjBC{$Dypc3YF*O&PfNXN-d7f zws-p`89II5R5#}70()eKZkx$x{}KI!wF&#p_@=xP58K<LDKMn4NNEX)K z^-apk`jN00Q!AS^b@TAGG}DyXzcz*SO$_c%yp)+!uTb?#DQ9w@yHL0yXW-kN>DuPeUFvJ_phR_=;_`r}gc`rK^GseNZ6)#}$>-86gjPa{Wh zbrTzU$9r>c>&)Vkc)xF%?Xy4LSsk5bimSHzMXwL2Tx)bEHrwRJlSuP<>K|Sc#?~b@ zF~<6117?beFOMc4bW|(0esa-j-RZ);pA1f2RWPF&i~WBiTMv7dF0_=FUr>zDW>CSp zea~llzU@vN7Q-P|q)l4hCMDA!vZnKgo#&rKDB$Soow%w97%=HH49)-Ndp7?4T-cxZ zFHd(yHdyx$(25D`_CYfOz7&iqoAL z`<3(ef~zPdaM-IfWRh=kvzmb4yBn8HBc77WT|QR7q);riO!hG}4C46Qf&UkL?*CcU zx!8`6YUyBS@S!qS^(^@ar4EC9tG@z@x9jusm&V@&l@OJi~HAxUk_baa?{Ao3ypgsb>tSGG^h)9Qcq`i$3l zPnUg;sr$|kY4V*nS>h>}Nv$2*<#n{4Dn2sPQ{fUqnnQe1&gI`%)Z+w=T_PtwM62mswX}y{(gFw*NfjGDN)W||+@y#1ID#)GzQutea3}aOcWwGATG5&CQ+(F^nyVuU55n~o* zgtVNGb+aJhKZETF+v7o>9f<4^IWByA!=nyr>GU6EV)NDEio6`!jg$Ouazc!y$%Uw> z;f0>D;5N=lJ`VQx8U15^QTTG@+8NU@(mZ|W9@Avexvq9!k;=>TfB33lZ-RX4XV!)a z-@f#_J?ER^H`C#95}M#=+;P6HNbOc#`Hs=3dXccoj^|Sc7ClFLZv$SHQeXZEb6WT;aP7 ztN)A0dmleiGdT4O5)LEDOF<{PD-Cxk!#n_sUp`I*eUQlzs7vTT>INe59RCkU*lkSrg2u* z+u6y6`3h|6OLUpxQjo8Ue;f=X;+)m`$3D)~?Nlvz?~_>cA(%cZQDU`_B=FAM9`ZAG z&-?Z8UAhXun8~rE1F3t2 z6HRjeGqwen+1!t3OlE8DagZ`_viPnf&Su*}!@k*zO<8r}5ksLeZZY34z07rBIP0`1 zwI3|>rwP4(kWB6P5qR7{vbajL@A6JI4v!$?z<=O?10Miy8gN(tKP&W1XG@i|tw6I7 z$R}X9+aijL;w^=*f(M8^ngQ6V;DHh-W&+-7P`d&+sev{;;x7O+B$v5pl2AJBet`c5RO5;&8?)t znSy;o#cMR?d~bsXD&8gwi*`WnzC6rxMq%JH!KZ}qQNEoVe$cSNk~&k|6<4057iI}w zA}scNv(f{5S!t+I7N%Rf=}%Ft#B$@Rkto`|fbdoWF_53Fhz)7fG(vuoGT`dY5goUD zxyhTZ-Pw68Z}Kn5TV@jbx4Pev%fQa&<#{9gXzMw=gUEK40L`+Ufr70&j5hPfXAR1x zf_2(U)oAEkJis~t+f(^-=VHm1$&lA)0WVc>H&LNB7S?6A=L_-XmM5u=gdZ)#F-_0u z41%a2K8s43BD+hhAcnui%Y*Xl9p~XcFVVST+|v#u{rdvniZvGlh~|h> z_(_8+M~Yt$_Pa7y0sQP~1o_Q8G)IGm;%A2>{DFYpdf@r5yx^~$@J~lz6wDS zm$>xc83$r%wqZ{RUZ;Q7Pmum?#Z{5V)~O2pD{BZYe6{DGwGfse6EaL-JF^IKjw&FU zf(M9A{$Vdj984BO7sjH%h}tsmW?ZCu$ovC^VLU3fV^I_`Gb^3&P?(p>I(}b5UW0*c z8IBm#;s;xTz}Qs;p0@qzPq%9)2uX#fjZn*qn;^x2>}Uf#>u_h4U;b5S*j&=PSK8N= z-EY?4?S}<3pU*(vCTBzBbu@)HBXdGEe<(WW6g2!;CeGD-^BYa*_LgT4r%VzQ- z<1o}UA$|N&V{-uOD)XI<^ijf57vJB(DJB7N?KlQXi$dqiXB4Nt1k^r@gbjKa2_3^v zn%mbRbAJkr`z`WJloPG8v?^uD-M&c=n75ns7}5Sa0{mJ^{UNHK8Ey4yAc& z&R#dV%oCKUWE)}E?a`}VufnyY*hZ~e;fFom(j(m31p?k(tqQ?*oemYLm-ZlJeu8Ok z9$?ZzUxu%gTm;p`!oB|ug5N#%)3;X#ZPX6f4jN7qGKJf0K*-kJ#7q_;Jx>5QR80uY zXB#UeQh1wjY-b#d7OlnIhr6*~A5K`Lr4Tbi6mZ9eGLD8;9Mi(C|B@N0{OPF3Em1aF zCxIpg;rh~7a<<#7e9>k)4D>uk6B-(~U?qF+DQ^HxX#}jRLkR?QP&>BrNA38mcZtL% zF;%SeUfPSXlVil%{Xeh#6LIeNC*oYEuBA$291~+t`rtv}oU?NTxL^mD}zjfQ#-IB|EUotIyaP7MSc+QWR;K}lK}#9e)OJm zqc4Y2hT=p6-h>>uSWO6IoBVaQ&jxodJQN4b`G&wB7qOnjtw2nmV-0_7NgkV2-Ym#& zb_8vBHV#np__`i>SBO53)0JocbkW#~nzD%af)VCkigI|IbcMyHjs(^G*uXMHGX|0} zW_2N_hNL%m{qK2n&K$Bgp<#5Q1*krXhlm`yb*a33DWK`=D#fsuU@v4*M+!1%c!-UU z!x^qP$pG93&ojRM$QU~_ATPwtlt7`LaT;W`z%TN*6d6imM4c}O2waEEh7~2RLea)Q zzk{}#-{|v>*(i{?WJdDbLB^pV8KqRdm%kv#g^UMKbKFo>5vJM;VJwME;5Imdo=Ae{ z`&Nh~R-uuZF+Q<<&c@73{E9Nf%=5y5t5Mm|wQGd+{iRE&IDgdkf0!ZO%?8%>4qQ<5M=lOy4A|pY^e#m4GqTM}8 zFUAEBNdZpbrF4V#TwS#M(z(LxN3A*sL2(dvee4j{o9iIS9roBcZ2a=TiJ3CDUv$vr|Bc;htjn) z5({Z1YmtV%IKHS5brax^$vwJm_D*OAjfSX(Fb-yFYzKB&4{D1S%ri zMA*Gyd|kV)Fr-(dIR@nH!kY3r*5j|b9-UC-OCc{{PAFQQIU~Qu&o&1O zryW5DF&hoUh$K{%0&*yf34%`Q!(eDPtIAL&vSb)HC~d{xS|N_lO7*+av2ok%3Vdz5 z=n_XP+-}gdKR5nr1a@1|0eu$9QC^zeQ1_cpM6I)bsB*iMyuGAwN!O$@?%x|+#ULU; zx&PCn1(@<3y-M+vFcwFMM>Mt*kqPC`ZRmtc3zb@ZMdtQ)=-{eJ0Ux@2LF|QiokT^z zSxsg-d6uq1;-wi^37$rBf}EqWn@&8gdvFQnXkj+mB&H}WV&WB?wuoDBij*MMw&8shEdrLZS7-ZoTlj3Alnf7S58 z>DdD4nN)cgre%<^?b%vL@3-wRVe23@ea~zrY&e2h5c4|JJ%Of+Ol4Gw zGQ*qR(V7Jm2-W9dt2C_MPj&Q8#4|AVS(juuVv`#c$>~Y1Az)yv-UmG!`RweEt9Wm^ zhe|KwnQauNul31DF=A$-C?RtX7$ae>LBWg;ir1q@`k!W3idEv4EMdrM=6 zy?UXCX+Z`V`Q>pOJ4?49de%De@MjQFa0eeQ*M3$iw3Z}0mcvXIRc-Gv0e(;Xt7TA0 zE3AG8bk~*AIKlV|<=UJSNA9fV@@C%?Ed0sLQeh1oxt3pEgsKub(irQC^bZeR9{^+@ zE&sm&SiiU>1@i?d7DiG+oQt68aW0-oks?~3W-NgOo%Rz#wr(6ZXxJ4Hh9^1^ITRub zs6C=SQRKenP&k(enORkX(DU+_HAsbz0(MV~Z9Vn~P5T22rsV?j6zS!~O(WxJNnf&h zC2`I6rjI-*IX*dZqV!Lgz}^yZ zLALk~2aa!gcmEv%z1{`Ip#xi16L>y%1Q`kP+`^xdXvH?b5e6x<1U&I3B~t*0x*Xaq z#`oV9o$z#fw3b9FK}W;LBI$CTq^=55KbQ@2vVnD;IaXseI&LXIvG|Hk{OLvXzq68S z;H7V_uYd%7bY>u1FF=b=AjTLwaV~_~HZSGWc~Ki`$};5f(f$DZekST>g19c^i*V$m zM%uf-Yt$}XQM^5A#b4C$*N|ohpz$w6g`54(yUzU}Rk!(ykpatq(zdu`dWd99!Vw9c z2L>&uj?Vt&pP+-euY#NMr>4j5%xhJT(cOQu1hT}KH+!nkswR-|wLAVwsjIKOd7{Bn;Dr9E!`)kznPoU}Kb}JUyk=_rKkE4y}f4Kw#YZ)T1rTQ3V z!$9(^uvy(F8yxB`PVL%H#^c%eYU=c@Va}>j_OB zcplQA{79uL_X?G*t;*jYv3rZEAUAM9w==PbjAc(aVM9v0#F5;$mtb1A&1}rNL#QD5 za$KFCQg#9s#i(J8IIvN00OU1klNdSb5{dGRlc%wCIy+6Ghis8UWVzu-iI}5)o9w66 z#?PQ}NA8okW5M1ge^G!+o^Foq8?>hSNJ$@VjNLm}ydKe~Ftd7nph}sTW1r8?F&Jw% z;s-QmvY+EAZ-VT{ABY@PY$!8DggGom;O>{iK99R-A^v`NBLCjx(IWGQO(y5)BFz4M z(6de+@}wI!z`U>7BqZbmnV14`k*!E6SWbA0JVrAD%;?o6m*`Nfbg1^Z3|iB3gQ!+r zBgAVLrgGixDsp*8|5q<(8rDRz@8PO6N!TNxpsb1rBWn}|WOE-y1CAgV0ilNh7uLus z0@|bFc$9Gv5e%{&l|>-LC<*~Y+XWE`Fv=Z*Ue@9!!9D>7CXdIRHoH|Biri3vu`EF{JY1Os0ciH5Itit@IQX*?VmAcA#u z!DcNOcXfY#(@q|9aC z6&R4$EUSp_rc@b9S_u_5LTx~qj8w(_AZi?08&x2B>m;5K>FTzcStqS!XzLIK~44$WyHB))NZqdZg={RLXBje&OO?~If~ zYp>QqQx!p>GiK%MpN6QuTn;&scvBp$jB6?i2gcR$nC*YU;IEoz+afJppF%9hl*?F|PjlzkyCmLeDy>||vCB9D%fEgya2!wW=Fp}Yn?|h>M zQu1FdBkaQ7R4)t@F?{y9sY+ZXb;;oeAHBZ=O{wM*X?<0p;h)i0sKTDjS$YDyXic*E z)jG+lsr{D zIeaN)Gb4g7lh>kpr~Z{_P<6 z{=g644o`{4;Yd7p%ZZvt_!&iD^{I^B;iUs=59*SLO7?TFJwW5dMX2IQKOZ8ylOlTd zAGhyW<^{xxL=Adnsh_2r_FU5e-fIcxB?;Gq$;w|$rS#wi4F%!n)AIk7Nij7vckwnA z6c-->NBKNDw=$t&jpV=&kw76Qay4Gu?^&&ldb6IRG||ZlZlq{Gp(2FtKpwY`~uzcFK<35iRIB!7AVzAI5#iG6;j=aHvRfm{CzV|9Uptvtq3r`Ufn zJ7q_)*9E`3I1$}GY$bsYOBRn!M-K1kW%-X(XqT6UiC>k?6YlCVWVSy|G)I(}b%W6F z&m261RBXFZ%;b5@<;}ioSJXIzDhZY{oG=fJ^=0oVFlIomp>diw-dAaBu6^NSS<4wz_SjChnD{&Eqd_g&G43dzOMM^?V%=GzN z`cN5j5y$n6DW!w&r$m13i8j8tpI;?UxUa?x1Ax>S3noofn615Xxq2<%ilRk|LY088 zkP{f@fXaPSi* z=|4_X*F>P5g5?CtE1Y+G?mOuQ>#f}HKG;GQ{rUrWPm!#?Nsi2(O^dejQz6b6LrGz} znm`vO5O*~z7vQU0X$Yme`f)Mtr(T{w=HjG_!zv;o!THdyJjIUq5)%|vAi8iPoWc(W z{A(~Hb9C$?=yDzmezJfJZYJKE$bhPJ70z1fSjEM_CyvBiq9NH>cQWQz||#4X?>~+=jf! zCCvTG$mvgqgRyf&Va#si>5LZ~{4n(;gsj{Rv@b#O;7STJM^<3p+nn{HLE^E< zoFMZ3wUs>>x4r6Vk-$@}TW0d0HZukmkG{ydwHcMrP}lQ0lp&E`=A_YsO+pr->%5V) zUo2?N2)}t-iCx|B6F{>gamCW*7~Z46EnX3(C-@5dl!WgwUK?Wc`HP8Ce>3u)5_>rA zcS?Q&QT!GRR-aVHLnnDuu7~%Go;8aO zs>$qNv?^WL`>|sXBN3OW$xMPQHPAd;=Z7k*s~luISkVR3^H8jx+|9~Gg;sY)rhp+5 zcNda-;(J+Ghpi+MG#@4gV^_Y|D}90)UR&P`ycoYj^;uB$G`-+rO8$9+k z0v@Aw%dznY7*rlYiYdV^smN@$_ChHp@Akp^tb^;`JR$~=m^V-i`M*N9DrYeCeejh! zqC$Jxgd+V-liZ`*Zaa7wQ+L*(ief0-U|B(!uQuU+fGSdw6vo$-~UIVVS5B{qjp@e^u5e-S&71AjPZSO7+Y!S z{M?Wmsws#|zpo)EEf@fy8dF%=uhkTmmi%c-+j`XoZ}LFf$;KZ;l$f7@^yB85+tGS> z^=l;*X)*8D&6Tc4XSRAhUf_hvJw-27(9>}WHls$vU^kL)*ByRtN~S8IiFrP2tt(jy z+TCEshVKy9{4)`Bt0CYT5TET>^D{AM{*1!aBt2GMy~+Rm{Y4*iOjfImhCpP4uTmbO ze};omVg^2kKsV2sO#6)@Bo$%HiW4T1RCYRfDJGAh*8`3E>T+Tnb=2XtnC z9G!K8&OCU&X$=_|(I_i?wMa~tNa?Rj#M#+hG21|dAC#)W!Ol3?e877K)<3euq;H4- z6Kl}nO*1o^gl%0LG5(=e$f_i2>@#o7g{(bQ>2I_gnF>e=cU$E#B{MDE(_d&g+|U|H zhTqm~dq`wBcQUwEMC^C?x>ySS=AyeYt_#?`I`2+DGgW5uHz;rxL4>ZG4n5DgSdy`h zCgg9ytb#7y+-Zn#`oE`O_Oqhwqu)E&;h|9an-N8t-{MLXtK>*VWkKyFCQFzDz|f3r_>8?K zY8bd|tr9*$;XhDasmQeUJ@+SF1cgRB_~E7zvzM7qv*D`h#&Cw&lO&gO41#&3I}6^5o6SE6D`OA16Kq|7E|?x1Ep zH2n!P#E8QC2aakBre7@4ciA)xA|@?Q_PrHC)3V(f0!LL;y0OAHxx(?|hKyZPzu1uCHU^0iTTRI|i*bRDKx?$u&4Q2TNZ{W3;$-6Smr z)m|nOH42B-$r8C_qU6!7d83WB8ALpHAYxw=bzg`{mES1dtvL(ym64apeLh}$ikM;I z_xF;VeH(b~j2bb-qIFI51Z$p`V$#`DzSTsm zIw|dD@)L)b2wXlVNFqt^?2X!VUhT&{@``kpC*l4S z*1JwdN#_Ws1!%#GGCdXPD6OuFj7dKgbJgqAZL07-@PPaq@tL6u1iYn-1dgRvmJ%s1 z$u>d?O?_Ax>|%@LWni10N8fSs<9m(;?!LoRf4yWjaKWo9`;m!%x{NPC$0F~k07oY}Y7`Su|F>{faafH@6-y5q6 z3b_SXS&9s|%3KY8&Ns}bGg}oj2$RW*jKAJS_WyVrzUxvsux>)ZBNetG4^1}dkSK!S z15nokF)-yZRIZ~Cb7&lwL=zQYjaPigj~w^;E_Hk<*|tNI{Aj+Hx)8VQvXPua`B%SP zCjv^$=gZ#3NnDI$gLE36UO`d&`#3j{SXf7(n^r#Mk>_k#KyjHR%3!eOpA@MWW0zIw z30~*D*QNxADbNMYER0(K4f6-$wK;j0_k}@y4Mox;o|r+k&*4lX3V&P1eGsvm^pH|B zC*bfBpq7mZx(RwM zKQzPUAJKwh0s>1bIJ?fT1LIS2)^ujhEG6?J3_+!n)G38D!*z5##%luO7n$Qln*Bg} zm^iD%ZjeF6RqO7?SWUdqi>Mm+RQS}wr3wmMi_=psPQwtnj?fFyt4#16hfw@M;XR_l z=J5ecs!eQN{K^0yyviabYc!VtqvIRAKv`!B;1&Rnej8!wyW?bdZ-6|mOA&UTazwp- zm4?w1igGBS$EjLXB3DPGSTZCR7>9Pz#3KEf8@?91sNQG@cPvnYsYBy=7|IDkoiJe$ z;9AIbc%QHT#IPB_I_HSP`gl}t&kbe#?$Vy>>5L}j4$p?|q*NEQ?^71s6;@;DRzzzE z-ckJ$2G5F*&1zX{lZ}&|ZR#kSqoA~HBXJl#)3o55`Qrv#w*jwDo(JdTpJ@_ZTE@zp zNM$*j39QqMbjR3R8R2?(Tv{HY(W6}r@R^qTs(6_HKsYoyvB7RR{@D>&q^^d=b#Kw4 z%d${|wYe7P8lZn^;U@dZd-Hth+$#&oy*g+!iFLl1H&sd;f_pEv#;Xv{tvcl7OmOk; z6cFPH$ZQ#k^ziX*l2IAXNBW)atw|LAIXbHuCT&d+Eg-z)sF~cKfXg?cEDLpp+sQ?R z-ee|1T-AZ{PIYpb-0%V|LCJ>;U)n6zEV!m(H@E^H~7`s{rrk1S%Pv8 zCzRf8Anz}vvYly?WWf0`U1{uSY9eMucLAB&z}(alDmyKdTn%jDxva9l2aog>-pk%l z(sgy7C>Z-v2e~)pJ9=-dB72@Y#ia>^>xb?hS7uCauLHubD5F8*y6vU8^n}l1p8bKk zvn82k2r*trwK-atTmx*pfKUL&CRS)jHHf53EW)HSE=_1%1-(5Hk6cF;ehbO&5y=Uw z4Ser{%mwJXye2;9*VFsv8f9&Gp1?&GXr$dx{^0*;gN62(wt|l#I=J>?&inR?g?cD{ z(?dpk%z+rE^aP>9?Nb{Iyo+(d(MPhV_M^wY0|hb&maZcXD*g4_LeS%jHiMG6`K-?W z{No%g^Ol(`UE+|BliNqM7mjJ)RaZkFp)7Z_ACcOBf!~flJg433VkkCT@dL}p>Hi(@ zi4OZe!`1(h3o24roWA_sg_jpqoP)v9xQ%BnrTuou&(cWJMj$|XW!E9k0cVY{%jVZN zTkW?We>9$a7Qy3a6j_Sc=wI%A>N%EYy+6M1QGZcPeNB5vTqEGjAYwZ12yl6pW<8U( zt;>5dw891je$;2g*Iy6eA1n9PCRAirF8%j7(|^Y$OBYH$_xfh*&h@$b7h3Oa>dHH1 za%|2T1#KpE2AoXn{Wd$V`5U6e$>D)mzYmv z=jo!#2Wehna~;nY-Fe&(b;Wo3!W>mBk7o)ach{r_Cv5tVaX@HbZqHGQ-1z`;BPq3swr^vzQMPORV~5p+o7LYXx0q7dx`tu{idEinlgh#-Zm#-?nX5Jd zSN3>kPWR-McD+4{HXIypZa#Uc>c z+*=VhQ9ttNaqpdYeP83o{g-|aYg)+Fv_2YrZ^_?U%@A34EpN8!O!nw7BzBjU2bkXBWp>0s|9&sRQMOsImZ}Bd@bZ;y<_b-lrWS-++wbfUl6mkuPzD@AvyaWA*@qwY|NQ?NGW~y?Vg0`y{_BZ~^)xlxMT(UG O{WxuP*}~t1pZpi1hV|h9 diff --git a/docs/README-unnamed-chunk-23-1.png b/docs/README-unnamed-chunk-23-1.png index 243f7cbae2a3102c1d0251a449eb03ef44bd2a15..77a7137ef1aaa5f4633efb1be79ace0d0e503731 100644 GIT binary patch literal 9328 zcmd6Nc{r5c+qgEAXq6C3B1?oU8B1j+gk-0bk$synS;mrzRCd`KvhO=1&oC;bsBDA5 zcu2x9JY%02%x`?YpWp9yz1RD`f4uMYz25hG<~-**=U&gbpELJ!&VA1FTgJMahxiV$ zu&{9I>uH&>u5(_RalsU;^8duaFuu_PrS-5Q#_XWi^nocQalM8Phyq~jGJ5T zty?9JA7kMxShxx{o|)l)Q(>8j#H3{wSQ4|sz%oc!2BRq?1o!ssm-KW}yb38EPQo(t zU&5FqyNN+!Fh-l38CW=j6u-;C1Tk56!3=!hP&LyY2YvLc{aILej_p2sIz?ABSXj=o z=xbfS{pigCCis58VrrDy(X>m8dw*nB=Nsye_bKA4Q)|CnNLB)vPC-r-u0i-udPD=*)dKON%hWtq(}A05=k99JT&3+bN#)6 z2jDm_QYSW%Xt+l3FffUDS(equpHKZs*k|?Jh}PdAsh%;w^_ne%ID0U4ata22iyxlu z!6bY|blY(ZrFNLeOp!#)rRW!b>JvXKM{PAnLLpj(IInCMxhB$#Wzf>mRgS_JQ#kj7 z2p3Ulu&!OL+|Am%aKT&(5E4Czn1S1X8u;cc3-VfU*r{(t$ZTu-l#C5Icc^mQvE*Gy zm>6ZG%S#nfw&{$DvL^l>EORPR&g`n|magJTvQJZFbvY&3acGi6=b33It>*Dp;pB64 zMn=neQI>Uif}#X8CCs;^vojxP&8+&J0d{LBmyZ(eaHo21uxWeGIH4M}&0g}(;oewg zo?M;=THh#Xbaq#_e|9rOAsKd~VU{+dV|tmv13=lZZ1U&%v4cABN9krXLcVC#>dh~? z>kSqSf>UhiTHGWw?qpXvtNk3Nz9z8{8reZ+kihh6Ex^90&?{ePT07vlCd#3(U(Nj+ zwTiE|{Aq=xrEN-P#-ZvE)8aI`tRYTz1~zqSM7jw+EJ~f3Qhy3H8YWOqwIY#rYv9^8 zhi_Gum3)r&$f$dYrsw&E%LjTesRxl&RmZz`W(Oy~KYQ)ES^Dv@MbymJR73MC`;)MA zl(>3_$@6Wq+O3OgfL;4EWsi{*j(9COx|Cmf)?v}BP&6_2` z_Oo@QK-?WlWu=?rgHu3QNMJ>i$B6TQxJ(6aIj**KaLG5OzRB|Z(KekQwGTayq@@Ln zZrfk*P~Y6;G28lj_)*TB_t1rKXggt@lBJuQtj!$Er+$=0Dplx|reC_X5lAYlUrPE# z3JX}#a8#f^F2(&Je}kc1fjgH@mbm*{y1&%LC6CX0 zb7_?Vw8L81b*0%c{t{Ev?wi`R(}}sPalPhcl}x%)74o|zu~}l+Kw|?JPLS)bYMvP$b|rMKcokkd{h)M+jqriAV_ z&2vKaC|BdXLpDajw~*aCZVh#7mM^ibB}i&qe?AVI5Sq5W9{yCHO_)DY^0QQxTTm|E z0HbN~Y82dQ?2|yJ2E`CBIe)xXkr2?*^X}9wni7{5pj|pw>6tH7*rN9HY|O#;sop6o zu4@P150rYed#rcN>jcKBF*e$y?ClShQJvCQ%I7L?vfPnIGf6hN;g0!ma9}neL}N@Cdy}b#?T>F$~1=XW#$6&mBTI?t~=ntn75HQ5qe5 z1x~Oyq#+#DzdJyo*v#K{S0}DYurx1{vsy^oeVOy)w&M&04?a?Fq8FX;UC>3wu60$# zR%QGHF7SGCH5sp}w{FRRzsU_`GyRxL|7<+sE5oQrjTGU0&+>RK>!MFSXE^L7CcU1j zo-e--yzcWT8D<1GQAev}^1Rv{9-T7d1|okhUQAV_c2ro36|D@nAg4dgU#2Z5rM6z9s03%exx3*+Ome@scl%fgNArHwW z+g5Bh^1d7Y1UlLz;J7h>UeDiJ z@!$f94z#rxMV1)9sE&c8Ihk&8O%%1jaA))l19&tlXRCsS1*?0{EV#m_CbmAQ$0ah>Bt(7IWjpR$Qg zy_Tu;cn2iD@9>!+SSJRlxni*UH zl~O_X?&b)UuW^X&;kY;VRV_y%RXj5vpdL=wyam3rKZ&oZfxdkC(#?O3_Tz@B^gsxu zwQo>u;l+{D(dBE#=yAvUU|bwGs4z+VcFgSDr*v+?dT}R9L+hwfxEs0Ut8tPvm@*!! zTS)iq@aHDuWsA)}NmzGJL*}2BK?AOp%AU;zrhE*b^wd#C_bc{GD!1Ta^G}DY*^G|4 zhj2fTpANHD0^>4IBf_{qrG0{6_jE@^_BWZx7r&euHJEe5vQoq7F^av_A@&;TKwAjy z#_GKlkcYFa?)o2vKJfF}2u=JQQ4-%F?0Vs*>{`MLnU9xrhdMek78M?9P~}0n9_`HP@^A86xE?B$%EB z@_Qbx7X>fNeOj<1mZNr*7bL-k+~?8^QAyPKUY(e*t0BiWu8Fi5ylmktD(m(3RBhbo z5=%zKT0EPdcjR)^Qa`F+W%6SFCP4W}$Xk1|^v03<3B0+EwllS;{d-@+8-)dxL2@N- z4yI;fGK8N15U8VLWjqegL(V4F6k?(V*~pdbCUJu&25*r2(W9>pRQ0{QupFU>&+yWB zQwWNOpjJ}WUR&NFgwXkBOV{?}%+6RENyf^}jjdt?E5tmU??y>ChxzLq-Ee=RHc&r6 zuHt(r$%)B!lqB~CJ(dU_TRlNo_3mO94tsUk(1>E;sjy9za8#0AThAk+-%`52WpRlc zF{WRJ&+sPXCu`1n72mYYs~oO8?TyF3vZPIvS^`NSMf?-9)kaoX$Mg-G`Z2fTwr{H) zF8YuPXKN?Z#}}|X_~6b%0=bkF=;U}sigEX$B*cRDh`g|?cvNpoSoxCV(=G)8G|lFb ziFJ-s=#Bof;K7f!l9()B(bU=+O5W_I<)-NecU zE%ZY{ac2gA8m#|;1lJ1}j|i$(lcAX?<2Ho%n3Pj5`fgTvim}Xo(ODZq{sLSALH&(}D0m*uTQ&;whn|%#D9o^9)`STVicO#9n?<)DcZAi0Uly%v&{3+_&qi1&qY}Xx) z&>zl*)-#NQ&)*rYhz)UH0sGFbD?2Fxx^B6UX7^Lhe)MCeSq$I1>3GvehaZ!WjRFx6 zd+Y2L0K*eLFcp;u?GMutY1v=yH)P+4yuXR4Ix_h-xT51{`OBWsI4Fg>Dv{SFBG7&B zqBy;$rIN2~9>JNcd3AQDnPECSpBvl@&VFCppM+42%E?!?p*fOHw_JHQ=I`CFkE&yO z=;2MN@~l)Vd|S|=4;lN^pM&=jV`nZWCkG;_%VsGpt%Vh2!jTN6;&;ePX9LsQs)8Ak zH&s9ULm!Up<;iiW7WHGVe~GISi67;RrVFM(E!%P?m-WB+pEl^U12QS!&f@MH*=~%8 zY$aFVxP^eYpr;1~L}wpyY3iqIqS$0GVM$s?^)?0<@6$y9Qrd%QvA;We2UW(371WKR z-dVftV)N1V!tfa|n^ms)z~6WSngeHft*tHXI?xCh42I_tFrm^`x@m85SDgV!n(v}0 zSIPH#fD+%M&3hHWeWu;ZiU`mXuZi;S)B-E!crW$mG}aH+2c0=siOX$V7tkewzvw~a z&>Hz(&6Q`fV^t2V^IL#r(!@gSlfms-;#}1U+t!CQDcNh)fnR&jR-Kwilm(Tv(C!r7112f}lEGc}{a_^iZH&H3R@*HzX(V-S z-}F$Zk6ZuSQuU%ur-AKR^Ppf?=VY>G*Fhp;?$_y3TtjWeW&I+wS%7nF=+?|VH0?+n zWM|T}Ayak693r$3D(ep6740DR`fGJ-phtzFwhBX)91cy6hx4S%Lgji)o)AK~GFO z*Som}^=WMnC<9U8Fe60y^y=Bi+HyEYnU+l4yvHC~@4j0KrXQF&ciPQ08nXY0n|Eyw zF+rqw+UBWWG@{9{uj#2HEL#;>haM(XHXv$}=3Tm|JaOIr&LwZoaHP8Qza)aUOk)+m zbDQ4w0I?Hx;6Xh+K1do{W$g1JTa$p8Kj9kntG^{L^|w@>Juqmtq*b6TN}UbY%`G)K z;~%t_Hxqi&etR~@%DvdLxq<(|I=mcW9miG5D6qS#oc2~e9R-1x;w?dcf=s|@w(xW* zE@)hl(Uaj^#(HYl9?`vxRA{M`c>&q$Bo0uBz0*M^xv~21W*z8IPdM$lpaCtHBS~Wm z^;KSY#brUkC+MeO6x8n}x!n|i-K=T&FucrCBydH_-v|o^#h{%T>z4)N3WHIT6OOcc z65d|1%VlNHzJ%F=xyhjCOJ)N#%bf0)UM>2+W*fTtXADAA$9TzV3t1@L@2*o;bEo&*Db)l&>mB!c`X-scsiLi=kDZt_jThnSHg zm*+1(09zH_RB}1;wE(t`a`#k&?;K7T5ysRue}$yE^QiQ8f$vg|ieB;;jL)y%N61Kk zDw>}M%@)K##}Z$7ti0G|J=%ganl(U%`!`G4lq8?Fl-zrhY_ptod^YujG#XThga zbX}r9{?Yt;#kIi}=8kuEYTC(#04_O?7M=pc;|e(Xl?ESv9^} z%2>)Wr+5hpSEnS3ftQ9{PFsDs$ne5!%6&>uECd77&?teSmRG`s+08D<@%-`PHiv~~}tyck*onF%)brI4I(oucWN#SAW4*5y!vN= z->APE*zhH~iP0SSy4Dm`uvPBtl7RLD>0?>)T1?N5B>0zptsfs<-!}VfZBy~erEj~e z3co~3x8B{kT)3j5VRPTB`^F2#$1Fpg_7`(*(4PTOpWsrjHzrSV4t z+vee1!SYjPG-}3H+pMvf?Y`8EEG-&#=}tuXQF|rs$3mdR>?#@@xi2vfc;Kw`4T{(gW&7%GELmSb|PqU|tH6K;0j&%_WD^FDvLM zO~~5H9acPZq5WFOoL4G%haC+Nx(;m1mRDjab)`wQ?(j}&R0mjIR)(MrtVuAg0iOweNbGhaj9HNH0hu0y?E_lvy8eQE-1*Nupave+UU2eD##6+OttE-vgy!I+Sr zNs^i3c_bBgc!8*T*9HFg;QL6VkVX1|)#8e%28vgvfa*2cIoCvv=n^EchO8C=?eWc^ zdL^jc04s(>*@B^?mt+?NMG@h}`(y`pySN4v_bafmgixU&n0v*8Uw_vXQUAA~PXF3%ox zHy+eO{rcwiP#ibwwP993Ktbc(?&yi5C|x-b#39&Q z*o|%%1cD2uRs4_f!5ofv=SPhH{IPhou6^~0lx{w3!&#w?xn<|?sFFAdB2PSWd$=j? z)B_5cYn9E8oy;PuAgK3aq?WF>eLC42&@a+tO)I1?Z;y@f#KDxDSpm9B%tOxs{oCLR z$mR2cuN6d9UC4+R2xfvS)?9Dny^qKMF4*x=90mQ$J7xJH82RMtev`eu>B2T*fl+Pd znhnqkr?=Z7U+8vU<)+L~K_6HT`m=|0U6$3Zr{n}H99ZtBZ5LQ{_JA;RE%%hyp;lRZ zZmoebk;=h$Hzq8Qudj>+-q-(!LgGzS-)QyJi)o8W!nwD&u)^p&;n26pWu7n+6cr9h zZLHRd3G4x6manqyZp(Wx+(8-U51dz!Cj2xf_2Jw_`%>u6pk(&yL$wb{mrj8c`%@cQ z24t_hEzA1M9h2&?+uV*sSXN)ufDqIj?Lo!_xMy8MouH(c1l4E#?6{+eno>*J2URKx z+4`hGY#?X3rTQ#$QzT|91r)_v#$D2*FQtY?Kq~*2<5c(lpLKzLl$qtOzR#7Ul3)2S zLxk^Tq>=oM?#}+F@2`xbdoQcV0ge|IQ_?~uW588gEqohVsHNl?(g(e*Wsyx-N zDg-XL@D}L-nQ@|Vz1^iQ8J}>CxuoqupU)u{v4bDr$B-)rz&$em0g1p9*8h=s z>`)5soqyYcQPOTy(-B4FDZ7C>dgj#B>$r6ZsLMr(wvJbxWsW1Vr}`qQCh4Mb{o*dA zZd2omh&!Pdh4x)Us)1ALVSmid{#BB1?;pBrO1v`#akAZ$f|XQfz^d-Qd}~ z6o0|A2NLa^DT0v6=|8{C7?gg$dik^4|olc%8M9lJ_1Xld*Sc$w>0vn9IU^nTc_-T9&N62TtD8x}tbN z=zetN;arZ3AEO_77^t@&wdd91x%K2&?@)5c%X5HSwDeo{^SL$R8>cl9&PVO99n;ae zZ@w)1?4PQL%f+shz`7kUzzF4EYa*xHd?m~!{c6yq*5_Ad5v_qO&?s))_Mo= zz+^3c4cnzqIzQow$39z_-F`v|Rm`?2W-YOmY&z93A3E}}J88j6IHolJD^p^E9Y*B| z_qWs%+X6dR^9+WsimUC&%Zr7Hd=Wn25iu`y_pUzt<_uebo#c;HHl$1FNVKp-iR{-w zbl~M+IppCBMID2bPN>boJV^Oo=19!N+(f=e4^SpRdfQJ-AKq9*eVx_nc4gYa97tPq zYT!22x}d;147j8^UG=?UFWJ<`Fa%A+1T#wzapn2NOtA^yYjYOHQIn?~+{q5>X?eYB zwTZdHwlr=Yn5kW~b~!yp&miUyqC34d&hCV-OQCRVV(wnhzi_bhY^CAAVSAeLU+Cv~ z$NAN!*ja_;-_lu4QG87NUF&hU#G8y&MRWP^A%mEhhallj{b`qCmm5X~_mg4I^!I1m zj4PR@oa7z`iflYvx{92%mpJW}rW0-+bH{|s{)Grr!F-YJ#l9U>dwuZzt8=}@OhG=N z21es%+0yd1@?3SVg=72DO015S7p6R%d5>ai_b$M$&{<%g>A(1C{r|gLpeFJym6#~f zwDhj#3Jz^1h0*XRm-l$I;eekQ-s)pGzp#yq)NS2!jTSC-B=PZr$81QWRhnAI>r+#u zzp>bf;pUo&;gEoe8m_~80k+>)(hEG7mp%PvjVf0M%gRl=Q_TwM;Oc4z*p|M=c6O-$ivh#kZJR)FD-^5&A)h{Z5b zvApXAqHWb|&4_;gCl}&K$Fg?d)u7zjP%O@7gs3Rh(q$9{-4BN%HWr15%ck7gM8awt zC72evgT`-KjUTO`KYSbf=yjB7$7q{tY2sA`s`sxDCY;gb;6auVvn zFngqjth*uyzTU+B~9MHjNI4UYQAMZ s=G~`MKJ)z=O77n~{Qt!>wLf7RKb9MSkKuy5H;44KjkPLo+*w= zf`HOnXyJg0kq{ts0)}$K_x;|w>#cYHcyGOX*ZS?V&t7}>%_SA_n3Rd6<&jlNlM~o=kMdF!F)7f`Xrb zK(LTd@SQv1YHBaFw30+cla)b?j*{I$$?nR@44!1=L#AX5<4MLaN@6k*lT2ik2YFsz zZ_LcfpFYL7gD~#Om}EwF{}&a8k%$ahMu8zRDhHSYBIe+rIW!bkSlF4BMNC#ECc6_c zjQn?J43fiiKs-2@YH2yZxE~Oc4>=f34A#TugI47DJBB^j{Pk=DK_DLP! z&xhBh_2$1YlpZs!6F)e9#(|bTEf%I5q533c1lAA5{Vuj8yi-YU>HFX>;u!9#L$QCz z0r023rat#(qF*PC8Ez|xpOH{IzFj)}Znt%JS`&e|T!wpC;BmZ}K7-!mGB}~2RI+>dN9vcx3uYWs`!@2+VRsW<~{rjs*>sEtPKfi;bx!K}3-T`wZb``D0aX;%Td!1jN9`5_z<0)+G78-Ft!L3RkkVn14 zu6)tMg~0z%U4p)}ox%F;F{3I>Vk_u$r103H^UkNiE>%w}_@s5X*=CS*P>xicSC84W zTkT{?YSYLbp@nX~!@iJPwFTp$U2ES%6t(lmH{fh4aj*HcZF>YlKYxw4ccFy8YD~{? zB;O^pHdGLb_Y0FBv~LYaXUO#y8sgIAt%nHE$q&YX_62OWONc&RKXule9nU$rkg2a; zV|jdb+RC`4?8Y$xrBUyPwF*Jpn>58cw7EAze%AB?(xw>0GRS|zk&borGkbI!55@(~-rMT5F zsgF0@)ZfV%M~g@#4b@1vEjrpl{}Nhu@+CI)UmBu1CbS;vB^^R!R?-?(T`s`PWt29L z5h^Dg@dFdS8V$DH7}wGXGt-5`h- zKJ)u4?(H?XSS)&I@r4_1~ou_*2%?+OkY{N1(4klDU7nPCp^Z^%Ut5uyGUS+MED1?aSAmm!IQ5 zHd%kSjxZ@H4H+DLk#TLixAx+d*m=is6nK2=vn@HtRnzFyF#f?bUgf)Xr}_5{4a+;j zsw=J;M?_gKrDapK*9EqUkmP2$OE&?g71n>p-8befq}B7H{AzCZ^=*zL%>U&%al^uQ zXBf3xz)AC+ZV7bBybfb6dQ>vCW*$W9CGGfH2P~x-cMb4IbtDAvKx}J5R+saZ!(063 zATZWXkBX-}6+;4$$aE+YDk&9I=M$G+1UZ&)n)E7qB7EKNQ~+g{fr`B<6`6>0Hw9#zh;!Ne9d4)Qc)68FbY-&V@rE z?Cx4e{*6w-z6{)ZMECE!i2;7Ck9aD?rVyfcMzl*xaJl-udI2y^q?)IkL9|jPt^FIxmN!^n4x(7qX*f^+wF2k6y z5K`aV#He=z^`hH_tC>nG_2?pq9xtj3U*Ito{(^eZcIs-=jRtU(qy+?EN1Kt}k~|JT=p;ch|f_@_vH4FnsxwRt1M9;1{kEihAyn@i6P&;k;4 zZqs^8(GGK2dZS6Ymr#cy?MUZ3J3H0;JZAW9LjC9<_JG6bR%T@qtVf6Ui{D0YWhVxp zR3|(l)`;&Jkbut&5OGMN1Gzq{?ndUg)fPC#kCoFdo)Tobb2zd$0o^uEE#iMx>RDSu z6c{GeHSGEDIb8FLW-4U%HJ*rv3|Ti&(yXUqwlL0 z5KX!5dnF;&p-#Gh2!YUn&J-SiYRGC=rNVp8)CLi!XnP7H&^heG-Qb%?fWIy*IKsU1 zy=Z3z?x($^x=+wQGZix_thFBj0t*%L85q zOJQ1@xM(46h~g0(z9Vvz-NEajVSFf76PZ<>%YD;pf7!Eh#f3}l5fC9R`?z{j^*<@{ zwO01sZwd$eN~nm_b*eh))LPek+6znF_;aZ%e|>9-_?>THeV)y9=j2Dma{D0_t+Zfk zB$!wQ>D{`8z%QR(pH(Z!UrFx=Ri!V)YqA*JETd{oNC1tWA>urlI~!4`P?o}Cob$&i z!>(0Wgdqk$3Z2EJ1w49f>h!qF1EidnwMd!X+gaRld^Q)&q&uwIlu?8G*=1q4Ql;hM zg4Zswq8hdQ&f0yZ^Z>p&nn1leGoLt6cRhvzY36$xCgB`19=mxiB{4uy(C^Md4cdKu z;7`5BOtx8{pP^$_9JToVHOSVVHs$gI zvCC>)^Ce}ZF2#y{kE{=v*P$rg1-6RCO=K1Iog;q5;3snmhnJnzM42_zIi&wF^4D4> zOK?N}MdH58ltw5IddNs*HS+7cLUq+*2JVHsgI#$&elt&?S~RBS!vkIMPSpHNF5BgO zKIdR#bhbk~Os2G-zaO3u(fYRadPW7cBeu^Pn?wUX+&rzfdp>2@%P!<}G&vY;zC0n# ziyp+E%V%^Izn;4U6O?L+2)y!r@qEgx1;-m4whVXNn^h#EL+}Fc-#R~A+`=Yp08L71 z{u}Xn23VrAWiLwSgS{qQY>q>(k5TZ1<91s>@ z-;u(=(TV|8wi!qsS2Wh%Np5U`&~U^*|6mv5>&w8;~D zIKo9^gqtz#Ox{~uO(e=xKFjggRdxP_)-`}PxpruN2X+}6k+L1Hcc&{KW;5C#GoWTm z!#G`DpiUu{(67zmoLJL{lh^TF7xYCXPZx0O>rcc{DcR>j1UXM2{^19@u&`80hXBvMi!SUN@))CUOW{IK>{8T)HSd)8Pz zVjwTI&JZ9e&+i2gw~RU$rVVe^;!CMtLJ}QMs&MuG!rr65t_KHa3_kjoSQM_pprmXp zjS8K~2pK1<6@75or~?<&$uxWhbAnxKqJ-zCF1$#s69yU;d~TdCCYIi~o0AxHEsc5_ zH|Gu*i8u*0J5)KuQK`n=oSnJ(7fxmG#zHw>hq-a!s)_gA-G>F zH41_W?uQPneEvlbCWqIao-aZQ!=*3-mEidsfe(lXG~ymw9Xm3Y!1v^)}%$l`5>uD`+o0TNmva+4i{!N?)! z&Cj%39_4f#l6WFJ%qOF+k*cf<+mv${9?d>+a(eOM3V@>?R}~eo77bP3+8o7GyDl88 z^b7x?>NdADZM8zw_8%=f?xT8zy{ceJj{0{v!S+4KP(_n9=`!?8oHh4zNI2Xk3B}9l zomNJ04zbt_?G>y?42tEwoANBx&8^6Hua90DqBr-3)hJ@YV2e$KQ5R>8mg6Q3M4M`l1)fMZjui$nm{r+FrL#*Ht_3;%ki~h2-5y?hw@fSVwK&NKG@idO z@^PR6?1KpKNU=ubEyFhB%V^Rs40sI6^((GnC`q_7te~5{s35}QV13oU=h?9@9z%Ck zi#y)Zzgn9PHq1i{Mm|`65u?$GwINqRZE4Kj^LFHil^H%sFbVq zuLdQde&xs~$Y|B6L-Sulr5cPb{%^-c+9|@UZSD2EM_W1o^Ei8*P@H z8|p%&QQs_o?tu;Go51}H?bT&tN^n=y4zP2;))F>By&x$|u0E8|+>>sbFmE)5&!V9| zR1{0x>%``SjTOw;QfeODYn%x{W5I$E4(qv8#ABd;4%waCL_nwuB!*t5thLSNh&ZX% z68jAht^}}0BltcofQ07lQtHyIHAKqTkA2wm2gkJ&FB*~5C@dQw>>Pfj#T>0OL(72N zaLuNY{3<+;ykRY7eXb;?tc}ozE$Fg59@v|pToGTliGPSR1FoqbBrEj^28Cqm;|Ivo zACCHt3RtO1XYso{ERlDbXl>fZN39~mBo)f>YR&J*F-E#d>RPLs7s*! zS#{sM8y!~;hMhV8maKMGqqL=O{Ys^)$3mgpYXq_PZ#dlxiHy%|M3pSv81P2?>uRfpfTQ1?dB*H&$MMF!hv zAux-BUwD> z78O1#1HBa|u!fk-^Us4;d8H9GbEPr4bZ!hrD!d=S(X zM>yI%&QeXRRS<>k$aSEjOVE`3v@TNI8Rqg~%rDza_b=0}alt5RgKwC76J)!J&E z-6l(e=FvXNSW`tc^^Y^Rwg9|b3H=?5-#1PIO4moFYSI1}&{$`BBeE#*j=+XA%Ge)2 zu;w__4Dd#q(&^SWybSttv5BW zNyz<=Bk2;rfi%T8LOa_rYvV2J1=($1SQO*L*_ucAn8OZ6{f9Fqf_@l$aoQiUXiI?M zOL_CMT9;k8$#L*5Ntxi0bBPb4Xod7Ai-|g}B+VTm;A}^P>jw0u6M6ZFNy&ng_5+~T z#)`cTokYEj9u=Zc2gpJlWMBX7{t84h=GbX}m2v5rO68EB$eNz3BsQ$Q!zHC#Nt2K7 z`baEjB2et_MHGwEnX55_v~Z7^b&X$cva#R8SI(#C0ICFeRWGbK3@Gh8syZvp%Udc> zvs^UdED4Ec`{)Ws49b=28k+`&GEXO`FaFix3?_X|gQp2t(tce5G1hgE$hTtv_iQnx z8jeCOCr^vN&!C<;LKfh9N_Fkvce($KH#z813!W;Sk@pnF#{ooJ2vpMFJ7ail>mGd( z!)PKhu_oH88~2NGwYs8QzqdGsy1m2>&$64-&Q+(-wchXG_p7MVR`AF=#nC&6$!;$_ zar=cDNdP_yOmF`P?a;6h@%0fgTk|0KO3{(47Oz#aU=KLqUEn>jfKsdMA-Ijr)SLzJma-Bx4GPC zPOA*r$RubWGQWI%*?PUaEM!D=x7(`_r|xp!O~d}_yc!%;xBIw`vI`SX@YG}cc#G;- zFD~*YcqcOmk%uxjS8Ov*H3kEU;x3*e?-vA>TtD}plUcWzy z0j`#1qGa1ut@tX&Ps<$P{@1ctfi}+xg45z?>us7=OiG5rApYIn)^u3cmzkDF=ogPn zBxZ_2Zlrd37U!|1>+yOW#}ls7h|d?38$j1PQ_5>f!~c zjoO`OjNh8d31=&YcVy=z^cX+Tp07}z(sbbSZYYmSv*Jt5yxG9|`9iM35HXiAm>7BI zq}j0=!Pw&f{J%sEsGh0yRz9;4$|81@EO7i+;3?myf&0qO`Z2Sa19{GB^Y7WD_lb}~ zrsB|{N2f2qSbZa@l=LFH&Jlnh?y)+=vj5~i;oG!?C}}Aet2ZmmhT%g}N(BuKMgGr( zzWi&Z+VS>5=;bHWi#Mq7rZ9Wh9 zpLWfNz#LiH^+S)>tV46VYux_=?3zofmlzJQ&c=XIW;sTM_5Xa+}h%QT^b1MwJzh~hBYzVViy08HQV;}xvr$z>px^iU(A=IeUWg-3?4ZG>qvdL`Cj0zX*Tv>1 zr{ZHe%6_>X|CXT{z$C)UonZW}^GvG5vtLIHWG?aLq$Y{~w+dngo%rtx-#%@ocDMb> zQ@5G^!cqENlIXV|8X??WqqcoUF8isKR&2%@?%!Ca#S+Wdxj^$8@02hz6Dq8B~;fWuWA70+6T|q{^vkyP6hpse>DFldpn=4a!Td zd~~1{f9>_?Mo!xtti5|xIT(r@5Bfc;lWwMO9NENBWny?-b|lx$9&HmOs9-#A{zb8o z$NZLv!e7%2EgW`s9AEWPJC+&S=6S=XV>V|>>k+B z>c6_mNdTuM#ILr%)qCCB2e$lho+r_87&@fq8#({3zWI;pkhXS{Q=%?x3)9JZ`sR^T z5n1|IDK><=b&2^M%fb--ht*eyl&z$a;iknUZ#8JC@>gqzElloRtOH^FxO`AT80)j$ zTh}c}mLuIp0O$@wRu#P8e$XqoD39%IMxPksc-k@PkPk}w8E4K;W9RIA@X{z3k<}A` z;_p}^?u9TWi>rs~eNf3k$qr)P5L7eRKIRDxmKeM|9td3HNm}!2`6H8{ya*$tu(N#y zftZ;O?*e3cPE8NSij}9n`3yyTJpOMSvH#!h7ewFRq3FDFYs!bca!06J8OG`6!C$hY z)x?xk3Y>OkTplgxuaK2MWPxOovvuuZY^d`2#IVbJgJTAep{P=TV7&dlu+d0!4ce1j zxB3=i$JTDf?8=x{7U$ax5VyfT_`QzSU^o%l;_?`gf1joa8Ty(vz-moq;9k} z{0OQKX>~3}HVW~w!Z`J69`r4VF3p^UH~pf`3R35w8H}C?v`aHG>Q-UH2fX9V!0ozF z+k(XKRG9jV&#SFcc5k%da1s4XVaPqT{mz5{OYMP8$%B>O)6WXRbCd+T!EfLNCt;lZ zYhue$P0+Qq;zLE%Q>PH96LK(x4ZB8CPH0R4jY3E0|5a6TS5M<1_A;S6uu## zA;Bq&uqR6RZcknL!#nsM0YL--!5TqC1VIE|$0AtQ!oQ0kSpOybU93f{;U!iCPLG9` z1#5Wgf9!vZ*uTubODue~j)m8?*6=p9wbr%p`Z-nvAyz~r)*4>^F=OFNEqslwh1UzQ z3$?Ke@cKDhUA@*Cq1IZYHWps2|MKB2{%y6j3$+XI@?87;+~Duuo1DB5E3y!4y-*7; zZ~_0=f1ALU=Y{9z#pdScTI=V9*yn}XXE^Eaaq#-NXwnXDIRZLJUe65yfh6_shN$2X z{2c*-20`Ji)Ccdh!*sjg$)3bKE$e>nS!bjPXmx(p{MAKuzR|n&TI{?%a}yIyE%n0Z zXD_Cj-m;ge1ah|$)IWK62~yX_&=0k+5loe8JdqI)R^ON+kRX8(O|gI|pSbZz5PJUq zLI}lZmMks$p+xL!q;)N^V-ob>siQR&OiZfOYNvSOczj6j=F$HCeF5}`*5#aw67jqv zf9rH>i}S|W^Ed(Bo!nL?7jcW1k>b$OzanfF^u#M{#t+uvy*=kal~H^|}4o0Uzrpr$&fI^j=BzO|>YHU@xN zd>_;Q-kTz(rDkU0&twom>z>6XD9P?_pyI=0-O|aF4s}?r>eKmTYyychA$Ur8e&xAl zPARcDe^Osx15a0}z-AiF|7?u{5DWkkUGB@-^70)o1UllrzI2Zm3FRwA4$BQGn8NjA z8M8h7;9k1%Si|y?T^cfZ=kPJ3jFHdMv!KTe0|6{$^a-j`|9Bv||5R_!YrQ~wU$fck zxtQ_rCuri!t>=j#;^`yT4z`j9v(fUETu_e&4p{l7NFd#i@bj+>h5XeB*J#w->9cQw z<1-r4GS)v1v8VvSjR@dTIZ2LVvF94nSSLJRyW8KSdlnRXC@1-u8Oo~_w@M~OH8rBd z7p=IKhdNHGhbAo;r5xjg4DkkG$|BfP>rOfPN=nP&2K2&N8OiDxKndSSJlzbzlO9%&u>?cSq`*JXo_DCHxN^==|r7uFX)o9_T9It+% zy!rlkLP|%9;84!Yx0C{OcX1wHL_O}CxMq4Rv#?`&9Zgf=KQ5oK^g22GNHoQ=sHS>D zJ)AacnM5WPI#$^Tosdd_NZL;59?G4PjPjvAo;Qs`I8$g#ZCI#M_o@Uj*_2oOEWruN zLqu`*?Cin>=JExCd8q#!(cM|X{5**z|55V8<9o{0Xm!DuP|ivEp+k+&B^IBEKU1oh zYBCW;y{chC_g{&fLja<9paXEP_oL~xC1M$#ZAn%ImyU6tA4|D%64;zx%*vVi0`n$Q zI4Y)-sL-(Xcz|W8;`*>R@?BEz$4rE`E-ZHl+IpOo z=eMAQ<$m#^w*xqM(X&8m;2%j<>Hl{4uZ%A(wjj$APLNwlPE-1IYM_uJZjX9U3FO}5 z*k5?D``*S6>5#VF9j&io@+qHV&P01U=W3SVZ>Znj^^wf3Y+3+(xs}Z0Mu7b_ z5+68SKe+~zHq_aw5<(zQi~|UQ)dT7m=}IoR#1W)fG+-W7m@R8zNT1#e5P0WqOPkOm zAX$HB0lobEL-LJ9ATJUEvStCaxS4#{x^AG0@)HPykObikA^-_Wf`)})LBoac1q?;t zLizs$LC*kIGl=?$pB3|i78l7U(EmE?6^=%q^7C=2B!XnvsQ^fLQglJinPLy6kLxZ? zMVm)4)oZj|2rqK*Z^mp&p+8XRkNr<`e%~+_dZS^z0oozAmKBvt@MTI$l&QYrCV?L> zqYoY}zJDaXpEy;sImILSg!T=?|K1j(NwCR}8m(K#6oF2j4OGW07QkyD21KFxoC7#k z%SXckBlfB71;EN7E@iCBS%`fqJmBKJjoZ}JrT(a+sMMv6%Hn@2BOd130>_P@?`z_2 zu0B5ZI{wpve^>Rqe$N~XjVWq!QTPunhoeu&!x_hiUXpm8pB&)kQq+^9BlNHxspVF> zBO}b>h5y&^Q1E6<#3&0zp%E-B>k;mZ2a>H z!^L)*=ksU7f4%rF$oq#Xn2{J4j2KRnD>+lm{9f^&8Bm}98}}-zBky|Wrx4^2fvN+c zL-SLJsi=Vb4QY!*GH^5!IG$DsGq@f94eV`_e`p7dAf6C*qBmx+lraqm;~GyU zLKuxdxJWh-0KG5^Mj#65Cn)!gd;}gA4+5C}#0?)M_~H!=~J0wbYE=ApIIaN`Mmc|!KTIUmqikM-4$inpa%j^f zsoP3{t^P`^8ClTNCn?j|NXw_@6-rAn;gA&>Jm+VBh1VIGQ1mrsxcqmogPc95mZ=oF^LX$equp8|+-y3vTgze-fzrC3ZM%@xkvS@Q?C&X?WYejP3k#fv^p z;g}qV^7}ty@v}lXIlL9}izG#y1sbk!AV9O9_Adnpc-g;B21NPz#+1(IgA5wVM}{UM zeNiDTr#<~0!@PD=ymxd#SNRJ>LYX#rBwevXnS(`**HOStdmLH7_8=)ru<7`%8&}c= z30s3V6=1s>3lNNkf9Ix}!pGWO^+L#InTsUk+aKyRcBtzq8*pTw1Pf&~=VCj>6Sv}n z7%JF}90e@JC4?)c=G>0=S&9qKJMke<$`lU+n8Da)3_s}vXkvo^&vsKB!@d%ubR0CX zUb_9Fq*DYo!zA4%4SryTP2A;&K3fE$FLy+=+`tU8ERX6Zc?5x>G9q&>pl8)2nSKxh z0;Z>$5{LxQr000}1}2^S-rXsbwjzFW8e19^H?+{;N)QHBB{JmZs()i^Useb|u5!a3 z49%6gb`=k%t%!OzgCPyFyD5pQfj7h#ME0 zng7&|{GyUan~Manpl8gCW{8!=#8C=oRZ5G!MdJuP;s0IHjyK!$nnw!86RvkVc!VKi zN?u$K1V_4BCM1IW`8)m0W3Vp$AX6pBQ4yQly#T7kyHR*s#jPe7g zsg+Y7=3DwonPco(>_xYkbFi&w#d)1KSO>jxa_wORuQ00n2d+kE`T8KSH*%<*Rp^|O z{&FjP$i-MAmfo zk#A4!m<>2(7swu_meJn;qEK2io7?^ZEW^fG7>fSm9mt<|>NV7_9}Pr4fem$ZOqtGY zHUrB-2dm_dP~?>_W~%0weSTx1F3(OHby{SIkEp`t3piM4*{_qQlKvQb1+qcnQ5n35 zzbECl4FLs`L|2XjaYi?O%$xY(S&RI%jlD3OG%s5kmNnI!B=n5f#cz7qSfwcJmoc2M zADhxS({DwvH9v<1^i)aW+`*Y)fNVdemzGN={xUb<6H@Ov04F(_5VLcAG?|gaGk95F z+4nNg$j6QHg8&`lP$Abs*%p0;=&!dpqqqAbqA2klj-8()2=1&fJN10$k70H+jF<;= zh^;sVXBBMU=SaSdchs_*D^@)ed%Nt^&Nz!y1HZYiX ztc@1lJ+K?VXl%`97wd}e9N26yQMQvVn6!lQ;wjg*`0H%^05fkPDglnhIV<8V;|jVq zS+8moo3TecVJt>HQbx^Xb@cs@bI%tS?VZ~$1>M`eC62M3ioOb3!L7n?r_a}&K-1@S z5d>EUtN7LF@*AUv@sPHfpa&Z5YQ}Dz_vyiab z;sE1YLa8Izw=PO~#*<&8`e066+2z|c1&dQ9jZ;19gCjf84~?4fENiNWsV}o(u!Fli zmeA?8?&O8_Gn;2zSAiDjjMn88Wb3@QcM6>m#U_XsX>3)Xkc0E|M9V%x(Cmcg4ucVy z=-uuT{_tVkBPqaca9H(*OQsFxJ8ZpB_(b4zrp65T`r~AaKd}=xrDWE zYk6$9Lpxp2++Z0WslDBw8edyuELmvi{y8{_udmFuX^$YFk{B{E)J9{S6HPeOtj|?z zm^Bc+1_#s!e9&SZU0FO(HFh_lSQBOp=tqD2FZNe49GG@IC z(}IrYMz*~3YCaf!YTCXiU}e9#Q;RVvmj#Oyj6yVSnQC%p5?8k|fQ|c&&4v(PI$9O52+|*KC&e{(`?=-n$?aN|oE}^V8 zj}zAKT~PH&|jkVG9Fd1aixzkQ8r!u3Y{l0>t2agJ6Evy zSkO5jCQ@?Z_Wg7 zA$tLjY#yJyZR5`!3&i!*S6fEQ0-*<-W@`M}s}<>H-7r@Ijv14gx4t`49@{gl_m`a& z0-+gKes;t%vP7ABrn}(@SHPR6n5<{HeEHAadK521_p?8P?BvQnuw9ais0+JoWIhjI zus#`TGfQ3sJ*hU4@KoNU0ZbI`BA;+ra!9K!30P$F)MqDF*FJ+*@1_%J7Oyk>$Fxil zi(i-AxS_Rc`N^$L7}dN%7Jr82j%0TWbRI{ z2bT4-=`|~F2jl2bK2?pRK0!9>Dq5(8XQ&Q;QhaNTp^^AQ-flM`#n$VJ?git}C`_r9 zdB)!E*vSivy03+S?_N{X4*6p6p4Sm=jg6=Oibyb+H|r$0TevdQt$VH-N+qS(GLGoA zfpB(Zx-$h02S|`A<2x!q@^`z*P8o0U3HKJFH2E0KFN6eJ1upEQi$U?Z=ZWsRm)p%nM^EKj;6L}8 z^S|ckqcVw;U{0OgA!MV59fQo;*xmPJ6l9_B5tJRR3ad@LrTsDAQM@jxbACx4)d3=> zToy>%8wv^y6^|^MVNRV$uELWP#lG^D*Z$Ip7-gtOJ(7spb0^?ch&DB9H`~pF!}?Sm zlR&W&UwM}M{hSh%r6J1yhaz#J%o$YQ!3&v>?^syz%5Jw4RY_&wFstz8mEmu! zz$>2yJ{6Z1P)ij0iLvA}~T zb=E%lMa1Duo%O>IlLTMUtmy{us*c2<*pq$cyQnC?DetUbdbIPJFCH!p^3Z};NFNEY zt!Xc`1gq+G$$Z{-PqbLfV{_QhhTz7ZhB*iLE8e9B_-5fb~1azXW_>$ zbFpn4O^krPWI1nKWqP*VJpO83u>@t}K|n}oDh$CfL8nqT_1(Qe<(Kv`47E~K{<4|V z&XdK)E!+w3wZfB)(c!WW;9SNQ!IilnIoiD&1OXb`N>!xcXLem(4(rQEs^nt<-l8ToUVHm-SJDSl!d15TTx7*(kd^a zkYP8RVD_yZ5=XyE@XJEU@t7^E%UDnEysGD;{+L>BPKsMGkXe4*klQxFlPBx@-!n|* zB^CSzoJXv7Jv5d}_C0=%Twa0qw&j+N z#Jbk83M_YG62;spC;mrk?*CZPe@eJ}x9J6UG*y>vp;5*mCFWTb$s%@p{0l7hi{`=1k`; z{Wi<3SjBIh<|sVkv${H|R!{AraDt90DNpNxmE%+ULlXuTZ}^dHgvuzA|8Sl2W6VK!W`@vJ|nVnMnAtK+TH-k#M zOmES-WfVumDijAZi4!FYR`7+taMjmXDIoO)+f;LmpFOvf;){K*D6gKa4?6Qz%Tn>8{tuV%dDbm7pN1WKKA3{98X+Xd%U^ zs139l>8kr-fX5dr#rNy?5f((F?;p7e0tA;jf4W>oC+knAH*4&u^eK#$6H_XmqP<_o zef*r#c2$*qH1#uq4S5yp?5JBz{$*7+*`bN4&a(bus6MHI^C9m3Yf_esWso3H&?5qp z8h4>Rh>G|)9ZA+ZRm8M2Bdi$MveVo)@|92xhHKj^WC( zzC3?Z#a%dPYBA+BNa}Xo4&xBd_so*_>fT{JSS`S=x(@o(sUh}*_dO%dz7`q$Wyd7w zYCPXT%MUWMW{U@jC}`(;%@>NTX`qR1RKT2&mphr;k10 zA9tnC$uVhUexLMWFg3;0$YsxQF-bFZAv-^Az_(@R;fAjy6g z{oSg6f0JJ3&iu(3%6f69MaxzD7PIRpMW|S&(yPyDPKjlIXC~_D^sx#RHRExxe6pd= zz?8bdcZ2n@n=o%+o8>qb(htj;e1a)_o3f?`opExb#Gs;oNp3>97o9TB%p*m6U+)0k zIMx-9$U3Z!j(7NQGi$&+_kHx+(PcJu8{u&45%5#P6Rih^4@S}#V`;J6P-+NZ^4I-PpWr2WfKfdTKbNY=S!{R^E1fQm}KW2*gr!TGm3 z)kCx=&=H6mn&vHEX zdsXMAviUr1DwEg?;!0K$WNj=e&ifL2XAjCKdXWQz)7=g>^H84o1sz)f&5szY7*o^A z3ul*@yDc^qiyG;Lxfv21IWCPeQ5me;X;=9S&xL}#vna#Kk%2;1XwWm{-5d=S62}xp z0s^X|%xqO|mERVx#{!Fy^vowDLXL$7ZH|+gE@xO@2t!$(rfb&9fA2p~T!|+(es3!@ z=A5T|#LuDx{F025*_9n$KRWFR%^)wa=Z7=$}@`!f2wiXMP0WTf{z{o562M_ATmB25AaLg)I_KICDvjDz9h7?xvZi9--S)j~Dk) zx2>y3aX|#sp!}{ld{G?BDSN!5uZaUcUKf3wVJz33BMrSQR%NIBn8&`e7a7{SpZrlP zS`s%CLbNyhT_hal61D6Q#=Z8H@hU4N{sUlJ2*YVgTi?cw?(22u>!`Q#2V${WY8L8j z?Powl6(c1^I~SlNyUQmVk--Jp(SFf+$GyFf=$ zQm48w(lXakef!;AHu&UlG*kRZ9 zOF!|zNrm<_VOVb1TM{;$kq&$(+5#XY$U%)yew`a zx-+YE$f87hntz$*Az0NklO>^&`-X4UGV+)mDQ{k+qaqhOL^_|*%I^mWL>jw{DHOY{ zkXS|04(rJm)A7rh+o51&lYIF{?}BvXe3#CXU!~G2Yd$eXD0M-0>utjmcof^kHbtf4 zSgHGJdojhkPhn#}yTg7hUTb1Ey2_RITR}i6Ynq9g=8yXTQ@X2v_Zs5Cm-(uzG&H3b7w2$OY4kX%`_1?{a^`Lde-YGsD;SN_|J>D22G%ksKu9VEK3fc_zH8JmX&s68b()#UQM+9ysb*TZ6w*tRZ)-=pGY~WRhQ?5&CB@rqlm?0 z)B>o#&q_YS%Ty zP4H%=VRfABr0EPNYc+ilY|pG8}wyW-cX`EE-%jvB+AJY_n_`g0pofEIbxSA@z+ zg7KhQr|j2n#mG*#zN7ICLP3|U*9M$6v>DQoRdonKDuNfdFLxw^p0*Qk&HG92l$SVc z+Kr0UgEwC}QeXxWk~+B2dJZSutB0{2ClF?Z=_HPnqG!z{)*v#TC$q;)o3=1p`#I!E zpDW#QCQ(dfUd_3OQgnSAyBn%_cOtHH?uZ%@?A}UL0Miw zk&d6SxnJAd*O_5=@G@QSup45My6|0Gf+?Wa~d4z_}7`%9$Z^ zzgA}q)v;@*9LbbJV);1@B?BR^SKR=D+xYqB#ZkS6F?kSS_f%_jrq2h>6eiYYZ|_13I<@ zW&?$kp5#gGWP=t2E5ng=Tddlsu=DD8eoFHuMS(0u&tnfVn7~wtxdQ6*#V@NInUn$7 zxZe$*4@1~_13YQD1KJ8Y!f35@It{s^2!uH z^=r~IJq3bo9z3?%3<;)-*1ifJy>iqwRbAr{^$1B3d9`eMc2wc@HIW)MNI`gR!9Tav zs=Y3dYBv)lN1Ck;D*Z!_;P>~=4DMNdP6+9hZb_H1PN}_}wMA)$ zcA|c@gTvZrN$W+hlYa4}FN*xZ(ul&!cLQOh6v_K%=dIym$#T`^WmOXWW>LTl%Mk07 zslo78$sZ%05M#O0E}oo~tuzl&3T|~ub}CiRFad7R@apO!lNr!Bowf+IPQ7vyg}Q z#ItwfQr+Aeu(IcJSjQ)&^@DEhgOC@^nWJMAVIobPVUH^FB<{ioijKDE?Xbeu)ApGU z6;(Xa_)25hZ5djzz^jtyWQ_h5iQVdbziDkt+MsU8Jbz5Z{f9}J!?V~av#R!-cT4(Y zFZty%gZHNbGI=l-a4nK41rkLmk7B#y0<#rZZQ}SxE${@eTEwqRlt@mwl%p->U z@KtA>Pd*U=DM(Ng zoPmVk{WP3jQe`)9c>2}0x4m>$uJ6P+@2AICN=~&@$!$W*_q&b zMl>Oy@w0Qz!$#~ZX!Gr-cAjjU_9+kYG$D+pj*4A~kXwEpLs5XrVMM|sAu#aqV6G~^ zDV{R}+rPE@*B_CsLwf>JEGlMeZo(RJHo|XesYfizZ1v@w(Ys;;5&L)nCHgW{)@9tD24_dHwxIJ&rq#vLQ`7sE?#$UNF^%n}ZOlcP zS41X>P^s(46mH``FU(HnHjZ6U@+Hc1qCalpsh6~|0Ka0iMco&}UxId5cK=`kDPcqn z(+Yvy6}2MxqAWoi5Qq3ybnDRo=lA>vC+KZ=!;|Q*VG+9xf0T$!a-i5__~dWVDSp~B z(sfsomIVA@#2CCxicAmMTB2Y9S`!6293Y7bM^y^NwTF%Kk~k&@<+*vMy}?Z;Wg7kc zl=mV@`vdLM-m&aJ(4PG3R$H(^2fHd)RBa*|V`8_d+ZqE-R1>t5QPVu)G=DDfO2^2N z5c5~-@jvt2-2l;{tabC1}%9JqS~ zzWdimazFL`1={56+e6IFtIv3nZa^u?p$VfNjp-|4K*Xf!H&w00x0qn@F*XS|CS2a6 z<~X#OfeF(#n@JKPcuJW=de6!Yg0iE&-fG=+-k)BOOpq8A(LLD#=~^&$?NByLs80ej zCs&3bS*xmr3nga-q?IvEEJbL5{BGi>gJXcz?=MzED2LV%(GnfUK(al`u&EtU3>uOH zEeFisyta9(bt5z<|9kdS#&0MkQ?wT1Hr7>Yq63@l$p?z8w@McYfu->`9(A(Y zP{6%<7nzdJygt-7SN)wFhbwn}DM#ZRXseu{sA2OTF%ot*XbZypL)j7FF9eVTg0OR0 zbmVa2B#pQ5(Ff4Bf=t;(R))|Z#^$jd#jL`_snG2r0|S%B$%rxImsDIu>-b#p z+&c@o5{f+PM5Ic&UCo?$N+aIwzw-l^7Z>^c8t?5T+|3?}Mi={?)+jFZUWgB<#7Kt0 zubRDQVEn! zz+r--E%~I5mJS%jSk&Ek6|lmx9j)=I`Agff68mRgK@Hw>ahLqI;{a78FF@3_^y0y* z)B9JzvSAx1{PZ_3PH{T*YaW7=w<>O=#9gh#TnLImZxNUXjht^vhrQ^HJloI5*Bf`F zVEsPx0WW&#Kx+$_hu&l*F1pf=>VV%)GdfYN-~{c6iyyH8g{yS}E79&thY#cC1@*8e z_edz7ZaaRk8;dffnF3ZBHuy$*YQ|$q_IrCfzy)gZ3;KwZ6>!8(aSiVM8K5Ep^t1$U znpfm^zDKS6(tTw`9%HCyKu#4Zv)_aqfUUN1^KmiZ@@0LIYDQbb9Jbm;{ahRqOpXCm zh=Q7lEkI~5%BRw>Cj1Asex)Oh5ir*?d}Ml-+iREexX--Su)fgS72ns(P~lUu$vxY< z%78N11g-2o-pVddgV^+Q^FG~{42t_s{mMj@2>1FszEgh5NkdDS{hyPzi2$vRz2SI9 z>^#vlDD3x7vo;;iq(#{vNGIS#w!@{=B*J6q=s5#*MWz1P?`4fk+dY`5IY!-dy?sXv zBQ?Kj%YHaO=9R3(K?$;XW^>zZZrC2MtxtLu?Ji&e(TIsySNXJQH6D5=Pij7*jD`9p5^FY_d1LE+H9|^ zNY>b-+F$o@Vqo)Pc6;aY6y0&<#YjF^yIG5BmHn$L44XjY$s9iPK843eY}~fMO}5aT z(R`hzqQ;kxvJ7Zqofs&iB91uiA^Dd4SJT!$78FcOMqMlBzg&W0qaH!b8V6NNv?o}y zygXEcdmqUy3*x6tyN!)zWn}n*9n>ONcKwmKN^LcB0~GZLZZESkd$wVz3slrdtUQO0E2wR=TTNeNoC@>ta;qqn&yQB-o{{YYtQ@!{=HvYx12VX4L1v z(DM?j3c6T6E)}gx$e&#u#%SNMumv*>r&7s#bc%y=InpTJyNM-*Sug^9Ia*Adu;IUOTo9#&m!& z|7yap@xs?;A0}f>ncw2bOD088((m+k90-K&ux;S)usWPiq%jfg{r7FagQj3P;MET=_YqrvEk%SO@>K7+uM zVwzOERvKAP_c9rvJZcHXNa(y9R>BR$EX0}en4HsdsB=Eg(!C!ZG?DI*G7)I6Lk5&b z`rwtBgCIs2ekY8rE4JG8dJ&tMub#%OC{lWqOwgKnIOw$tzIIj$`=L+O9HK=DWe1fW zVUJpEt!F!Ok_XV9?CHv0+GCIxThz=PAATkXCO5<%Rj$U~4d-9Cbzw*MKkKea%x_vP z-}Y|z3RJ}G(0AH8WSoh&2RRA(cv#)pdFq+hhy`$>Ntqt%k$Ov>@YLt!v1-ZPk_da% z+DJfDEDSz*hB#}mP-(G-D8QI;`KY8}?_N2Qu}?CYT}<&ZXDcl>$_9s#`p| z*z^aa??(#qqX-D~DA5e@M><=gv^)LNH8BGAfhbLh%!R2rMDGy5Q5~e$L?xfO5tzrl zUNM}HyG?95-C%+-Ew<{9r8lof2gU}~w1aiQFTb~;2hoB)4jBsm)?GDcTxOgWgc~J{ zR*N1$3jz<3q>Y;8!cLFbkmX3eui?#VKZHUVd3a++EGx9_ueoQe=@+wfZv}EVfgA~> zp;bFMgmZy$PVp~_-bc#Jrk)Z5endLarT3bhzu#_$eP?6|4Pz(L+wr^zg>Le=fL>QA zkJXJz9^2v$h(BRXCc=dyTj&UFeq=0eu_B|;%3`N%*#kr0t&@CIprHC(zB+ecG?Fam z9}DHqby7A#GZgvyDh+DJ@1?iQqGJ1$gRx=sM$gmXNgB4sO+`(`MM|7_+T>N+9GUV~ z?vA|CvEdEu?BnYLu9$Dd8KI}cuV|<8bUHb2d5}LM0^FdbCtp4#Vqq)kC$jg!PKq2S zP0)ryNg@1%He%D?M8~%!W8V#n*VCmgK01M+j$9Pn+G902QY<1?n$m8NHZr`V7r@W} z#=IYCKcBuXD$JqU3$56|6ULiqgcoD`46|ncyv`q#=veu*>5TxM`}k3+R>+sd?lg8& z!tq4Of$6<+i@vS(JHQ;S)s5Wy8To z(Z0n%VabOad+d7R0=*?nvWYRNJ;K3LW|FaTK_M7H! z+AE=n?CgyCgMZ%#aC1D9W;>_S?bbU(FU1f~!6-hfOii<>wD#KVoyW4a$g=p%tyQ#X_pN6Ku4HxdRLX*z9df*vRJ7sFs>+~ z4fiM-eI2LFlf;vCiR$k+rg%McDDfD=#(wY=@6Y{ZD)Ep0FXTjhjy^$Yk`7QbkIp2W zgMJX#Yv?rsXegB^28s^fQM9ypbU0-3bF&&^w)pVuPux3j0Nd*r<0S7j(fq$`@D~jG zi6k**kcPF0_dhh06#2EutCwwkf-d&kj>TSvCMYAj(WOhG1}nt0e~W^^$xPoEw>pd? zLY~s%nT8cl@sdOv9LL>4n^5rYWSWt6Ij8{V%u$*r-(?kp%IvNYl)7bL*s-0aH%dsv zuAhJmf+UDZXh0Mec~%Nf_x?BVOylZ-b`u`qhuaI{7H_oX7aCMlVQua_H!^;7?j~Zu zWvoK1o`*S9G%Ss{KCzp9Bp}MJd=}TkB>)f{@|OI|Z51;fNe;uAF44m!G9b9RO9(Bf z1q`L(ZkHmtL-}+Lr4cs6=v=`AF%rqEL5+gy)wrd3?NFw zPYFq+s%5K{WzX5x})F64EnIYkQ^MGCRp8$Y{fU^p}thp6Wt-&Q~yZ=b`X9PXVb_@o` zd9`|v>ppyg2N@h%Pz^BAfdj@;MyaWFn7@<2F$UqZrGMuKo}wOiZ}O}D$x}dX4d4r1 zN?mHf4f&6}exmw?ml>_e{{e^40AJ^80>hvP8_bsyCo`RbGlY465H^^Sp;{hi55vPp z8}F{Whn-#LwEn4l`*dI;xyu17n3wmzuYb0v{iox@}7L-)qnZFRG?>KDr#dhk|`dakp55L zsVcFHIsnrmk~5E4XXKy0oBSw&SD|6tn@FIpF3j(5I_Rxl8bM^awv!g9`k5MWqD@D* z>%_mQxKom~r&&1rF{JK@@O>8UUA&SP81t|Nx_xapjC0K1|AV$g;kH0O_jV@h#gOn; ztz2b9nGAiB+mKU?<$sGnBq$s(_dkQ+zYj1)(P1^^7Ic^rQYra(P`7#h1!&y&xTfOXISo3}zVtW~xH3`I2LCEv$@qd0QQ;c2eB!^A~SI7SXmG7*Nm-M$PtDI&Hf!AG9#R*M@jW z6MP@+f!F-_h`_6DYaRG=bJo*EPM5$erAh&`0U7MC$}8`+tSq^NxtPcpmgTd^%#UQeq*$ctrRo^Z@?+ zv_4|yHhP7B=Wr#+J30QJzP@)OVPfXoyIWt6Jl+b#PdYvf(WI}f_^Sw!4=J|=^{IQF zA6+(ytIftw=6Nz4`)@ZuX!r0mH*h~!14L`xv(#`|p3z+!$Sv|6xfZ?kvnJeKG?3UHb3hgPpmTMf0zCG~O5 zH{;{Gm?9Nd1TfuL+Xu;OMZLpy6K5zM1=laEWRNykrkARXE3W9*Kz<{Nmu|Zl6lPl_k zY7?>HslNFUS(37($XT9YckJT2&#R3naz@UNOM9&%IWn~bjxSF;y*qsM3K;XAJk;)pdf09iTzK= zx0scqp7oux~%d!44MqhDq`PKx&x_i(h&f5*Mcf8V9_|1A2qo{=JVH-49h Uhoi&)e*r<^o$A{%Y17aD8=wcO-~a#s literal 19252 zcmdqJgO}8QoNl zNsu5vKHulQ_rCwYec$`sJ+phB%*^g2XZP$(c4s!?qlOX*AuS;m78Z%}2l>xfSU4RI z8H)ht0kep_V|cjjepJ&|c(`L>iDO~eV2O)ki9eKaST>M{&*E4%|04cPHsUr95GVeC zk9z>P%>(y;$p0b4{Y(5e#XVd$aStWL=79$Sv4K34_kLJdapDgk9%u6a|Dne{B*?=R z2YD#rad1c+{Gq(h|M(GNg9Wh>hr~UA&A;>qj{jr`1P*~ez&+&tzRAz8FFhR|Ck~IZ zfkPhPp@IJp;ScE^et*B%+IkPMxrfKy!y)$%sDCRxl=lifo_r6#^$GBUfh!gkb>_eM zk+OSG5*F4gEM@uk+Fn@)*$zSULMcUcjg6Eo7JU2WkonG?LSNfZhEi&uZD-&#_1E9E zE5FAMzTr{$5;tPELQ?9FilMu$mBtkM$}Awt$C)m4pL+L{5y0{|_HI zHCE?8=2*D@e?d5EBg3e|U}uE_Z8G%B_roB}uwH5yY~>_XVDqMVn5nKEkUP1HlqmOM}KyD`vdvI@+xL z`scw&tjrBl(t1SL5o1G3iBsrTcp+;22qs)nF|#_4dVlgR`NPH4naF*y`r2bF5yg@U zsS3c8Dj#RdhUSmtm-_~0$cEhQJ5haK84vY)xwQ%=-BRnK6K>0@O(^fQ(aYN+bn|5_ zmHcO$Mt58fiXG|uRF4!#?@{*p`eLT)hP+&T&f}icS#oN}f6TR@d0`NfvrpSF=TCM& z5!(xMMRnij2Ez^I8XANrW?~FPG!|ZJMR+}J{x$V7bIlDP_qs;worWA10dWQi!D1-f z^R9qqMh7kpcb|{NdFgSBUfo;ChKzb>q%*Byf13N$4Fm;08fmS6xM^Co+m1HZUMe)u&zW&*HEURXG*zjFEl*k-d? z#|Dq)s1xk>wOSzeKwf19h4oX`P6tQsRbjb7p(&8rf_lYeMLVCMaQW-+c21+*4^L>M z8^O2Dj*CBxk9)Y4e000lxNRT4>i?CvDjTmD0lh!xbF-6CuB#)7S`j$r3f3w$ zXuK}M#B+QqgrK$4$?2|>nRanUQl{}3sqAw64oIxp$cBY-T!^m+bEQ85{+KLiP8es> zCKTfdhaTZTIy7GYQ7Hdx!9(% z>^E{xMU~70a5s}Oc&qi&A3g9o5iI@@m>X@@SWpzQBHj1@`nC8Yo9QBt6Dv&yjxToV z5EzHDM^&9j&1qXFR0S)GvyA8lqiNJSTrzhWCt)0LU_scI+eR48k`m1uhlCjuj6>`eYw!l(gt1WPZ{D(-< zLeeK{dZQXq|B7%RS~MFwUa2ZhUygjwC|A)Dc7UfwhA_QZL4bsoFO?BYdq>l`Q72X1 zr?@eu5gTl>M%{1SKbV8$^7*ZJ{(Uh_kPVx(-vCK2LP7T{d#Y>M5E(>frrQy@uMmRv zD2j7Br(^c9C#n6&k2d+Q{|YDA2Yk>edSXK|5}pQUJ-B>t)z%A}JHoIwuhn<3GH${>j& zDml+Tj8Q8AbH6cGh}(}#Js4F_oCCWcmnY$Kdn>%RHqe-KFYJa%wXH4=Uy}vpH{WR4 zb#~)A0#LuZUmFC~SrTFUF8a6tnGVU$4d}6i2Og#woIDJ$NkM@TSoL2y=+CrG)m;)Mpe*G;OEP=J*sBMmg^)SVK zH-E*46$tLYBEZMSc}4xdgGejIzQ5VBjn$OItRVu*(z&o+{UZfN___049*5aK!NHQ% zrnsEmP=-fpT}t7t93r;rKjUKs|7JPf8-)3q#qxr#*MTSMg})wQG1))y^Y=nzTu8Ty zN|q)GV!e`Q0VAZ{aSrzeJp;-Ug&zNeK1l(I052yv)+l-5B!D=X9ANO}FsXBZm`Xj8sVhKp}b7>RKO z8Z={J?T2UF-Q2~R^-+<@xb}{dQ-5clI@_9l>?C>nL2Rhy84)%PUx@wPA}RA~;Ty$Fe43SI7BKf`9AFpa5)+ zBmt!0NBV&@hmWOc{0RFV1YMN~2TMnh)%e1ff8c^2rR5slEoO{%`(Vz2Cx8d~e94C9 z-u{;1iioDBjjTD=X3^$tM`%s)C&33H9=lg(7*9=NtC#IoQWXD6!CTZI`2WPX_+J{1 z|Kno+q1B^wNVTzk#TVE8qJN>#@F@Xt_djsoWl}lwS3%#)w+&P z?*M$ghw-?f-1~pOMdkZUd<37LeI8X`F}!PSmu$U2U7i+el2RdnwGZvc&&OR35m#=E zqE`GJmNCbx$Ki=mxcGuS0t;suPxMZ=4zxs&?d^VBN{VQ5>NvHU<1FGz*6Omdd(!j= zzUPzllFUI*!JX(d9^^c=F@W@jz}5U?8118TW8z*8!8vMo%DfTl=O)9szFU&VE2_E= zt;|jFAE^=N?5UKqH9jMH{_%`{%z#=FRzIdWvSPShU~`bL|D$;%JE*gfjy={Yu3i;F zb?8|1#+|fayXRJ{k;~+}W!ItXn%$zyTNUqnd}eHcC+URIsup$Crt@aL55627z_id zM;Z^;*nu7XHdsYiEi}s~us5SQR98&GkUn==#5Pj+gChYg*m`x&v82Uey~YudsSyIt zm+LxT+&0;eWMuD)Z~}E!&>O%J!_UU6%4C7V;aw8KjrlS|M&A$i8#sXe^B%8VV%jpk z_h=m%|K)F#%v9H!%=KNC>d-~QpL?+mcS{(*uIh^HOzf)C`FZYvA?Xw)Q{yYJOkl_{ zKk86rj#iodkloTrK4<08_C-aqQbbmQdWOcHRbkNMHA_h5#4yk-8W}?Q1b8$|73ha4N0J!L)meQk6 zp`&>NnJ<2jVTl(V!$Y){D+@Q;uXJSMX4v}^5BRLyiM6{doG+pUCS<|XDtTM{+;WEw#^V3Z<%AO0IxBW0_ zjF(a?mL#EV^^hBRf;;25OJ0RQ*>4Y`W(m9c_q)E4HYw0AXa6PlGLKa97&g`8zJVyR z>hY$b?;S| zQ?&4yo5T8 zDC=F|2tYrqvKbVGy$-Q#@BD#BYT@(>ILXk^0S^fSwlk8e za@b~btOJvejqCwdIi3^mMWI&-_{Go2-|*v-;v!Z09778<*^-(D=;sP#FSE_7cx3O* zjKXU@B5AXw+vft6uEBG_ZR%)NMzyrkJ`-@KZwC? zRv&%7KI?<5GF@n~*UT=~!QJ!Qb@p{Ow(ru%jNb*ZB?BFpuJ-XruQM|Vwi(`Lf z`NLg)`A^50NQYpJ{^N+b-$k7|=flwKs=jRW%;~Is&Z@HaA~HYFMSlj#O(Y3ore;rt zXVz3)XlPjMe91Jtvx!{gHtex9rSzZN{!>IvfB7Ya+_5gM(}>awooP9EJIt%~in#8A zj8&XFgcN3WnEI$@)p!t5aO$ZxERyE#f1$!Ow;9C@9QC7&?H{h>P41r}te1kQL`K1E z(S8duW^Q`Zh~6vdYGT$67H(21J1}`GF`TN=iZp@&REOr|3~5;VweY8{Ue;BA(rhMy z+8ff~eCeJzRj}RIn$wLD(GTz7u4mvm^R&D}^^VIaWiL=KAl^;??gTpsafF#=|Gop! zzrt1hAVx3LC=1q8itAuN5i>sPSE-QQT>nR0CcP$6gJU#%tx7}u*KxvKb4l+yiSu)= zc^ZP7-Ec7Eay`CKVNgJ;d3YbVZUt0LlgluSeFoqjt? z(8$5>TK%s_Vy}B1Z0sK6c`DPt@8(zhKzpRiCLwib&*pF5C9<|4J000svC+*ch$daX zbmkX2HRoDia0QjCwV4mgjCQsnsE(g&I7V&QC%LAsSkJEiwI0CBJNQWJ_Zr}*f6E^w z(A8E@-`j_YSMvI7sm^v(x~26JCeZUZjD%=Wjc&|FqmWb4j8oz4g|D8BWJ)r304h0K zb)Bc+s_Q7Nl|ywIk3nVS13?PlczbkjFleyv%vR>WaQkyrbN08#qw`ICpIOZ0^OPD` zHqKJ3ZlXf7X3iR*gR(lJX9?w!Ai7(Vr!@_m$70&J7hgUvDY_Q3! z70xNG-V@}7%q_~6p16wFg}1w6RB^{{2JvoI*F9Et5p31+E3P5`JfXI_m7m3u%1++FS*=&ql?ur zqqA9~0YayuN@aqH1IJe4iK|SGIP7ocsm6dMd>h=ZFOsfmFM1X_&XQdjKB-%4+H{u4n zR2yXdrwDE;QLz~#T-z7dJuC61q3zPGGMEIn&&aItnyu^x_tX8OQ}nKJA1H!*!Z1d{ zH-7$_Dfy1u%s>{wdHq_V+xsaenBaP|F2tJ4 z!z^^Mf;|#O>>1{aICT7BzhB&Kj!|{~MWj_XU@8?DX|sFJ9|V8?*GRDO32?(?VuPE= zc#KrLikc$4zbU_s!}kKpXt?~{=b(L!lh+waeCfdA{{nqcWPeF=NjJAv`J?kr4ytdm z_eWWK2t|wz9b0R?2tYBp1#IBp$8am}dK=H1(v?v|+D`s@Ps&*KolLl;S-qU!DCZ4E zZkT(!pD&0P0V1t{0hJ~M=-LnGwTD)din_(d?ARg=VUbl~AW-w--7vBAdm_0C@5SIY zG?`&pej1u}$lbGk1w(iE&9kUodwo-`UYM;%oGFbzZP`x{J%I7#jGT>;DJJ*ZK74z# ze_H(yR?eDA`S%E=-O`Tf1Apa7&U;PET}LlE19weE9}BllawAp_9ZPh6A%5sq~ELq-mmM_i75kjUIl>dLQS_;^8e2 z1XoZq&?y6Sn__Tu-V~}SzeJT<`>_xoEjFD4;7jk!n>mdtyG-kB52U@xD~dXo*8+!-o*{dfbFnUVv6i|Mzm?(p>QxCUda8%gR^F`maP;&%=QFFE3K(eV#^x!IG z?N|`RsSzZHe};00m_4y1xYfDfD{r`WofHx1u4#@y>xHc7WJ+3Wc$K19>3^^3|7az? zg0JkVUaF>y3&iYHOYE~#f0w|w-M3@?hg$OHlS$B$vaax)QTH{(1Zs~L(A-ftcD6(& z>5;^98lySBgD|P+Uz3h0HcVz^A>A{K@x|As{|N(l`3{JuL~2}e8)6Im(3KANAVFI? z>pnJ^w3yA^gRQjPPfL?uuk@^}W?t3h9)#Z2lx_cY@W^pDl-yk%jy|fY(NZ)!uy;75 zyr}8Fmt`JS6?eQcY79Q0&1%C`Ljl#}Y^M@^_JqEk%-Q7!Re6#$T*tDc(L%tY6dNkQ zaZ=mbwt)#*$<|I&kewV7%`#}_A=P_&eX39)c*6~8C{!WvAa3X9t>M_Ab>RLgc4yo$ zK}z+$3dFwg<~;3+x-raRWmDt2j3-H)^e!RMz%*5dsSzX;1XbT&jLjkTeHne0dMqrt z8*C0Uhqqah>Ajn4OV48vCc~}@5P-c=bcJ2K;+|iAdg1J>2AYuhC~4Y~ zybB#&p70>OTUM)Yur41@A}=S=r9j@n*t0D|-l{D{|9K2ncY?9`Q8^q;A?MqNl~7!( z+U~}yC(K4j{c+Ym@5kg2=m!-d(air6C)~{kZl`$GFpmcaO&vmY_>c%O`cA=QCV?-yH)l-{`$BqV`*8gZgQ| zXix8K49TxP8a+bz11ENu6ms@Vb&v&l2Xi|Hw;)K7&80+4ZbX30jaHK9ecHtwrH(zr zsRvhRz<{P=sfA11!MDfbG50~8Fj`#=+487OU46{i@e-65(_-U~KC!htp%ACNw!y%M z&c7dx?IzO&9A#bju#)9du*H=|=LQ`LFYhpGWPEZuqCfqyCx2Y?Y>n2bh zc1N#L7_}FCv*3H3(ePTR{7yzrRS5>TKQfs!8${;c&`==r-Go9@AHDCh>eLc7uPoyx zk+DlIDh;!by!JQu2yy~9pfDZ#yw$&UIPgsGa`l`s6paf9PaZ z_-I_ccdr>^?QmgNbCTCfVva^U*ku$SRJ(kueFdNNeL7N@dE0DhRg4k>L_HzK77(ITYegr9lnn_y9bE+)oEXt#7=F-1RpJa;ds55KZog=0The;2O- zqbHPh01)I#rjvY7O;ridb@=s*WCuWfqbf$C6WNZz_sHA^G!AXDJl^8+C(+d1F7W^#14<#|Ps3QW%dS7PU%}W6T^;25o#d(ZAfd z1TQs+)b@3mZ+fMfoeT4Yn?Dr;+`Mbooz8ipK~j{OtMAS0J_q7Cr!+n4AIb4#OpEGT z#(r_~?IzXWr#OD%-StW;ikXE4Kh{5G!QnFl1vCDh=gdU^IqBYO+Y$+6i1}WDS8ZE| zV*GlDIY#mrqQPw7g^$@Ad=Cm|M(aU8U-N(F6ipZO&@J7Zyj6Ejay%*HTr2yj zzg+Yr)Z91dAN+%|8ty-x=`8xm+_{e-5hEc+H!E&PJm!~&z0c1tz4tFgU`s)ZgG6U? z)$^*iORQ78GyxUHiEL5Y29X<1+Jq%U98wW2Swm%FhRw}a6|QdV>>I|8SCT-i`-im_ zY5Eu~HNmuV&+UsKtaN`N5nOmlON%-xwQdE;+lKmj^u^Ox6n~E-nz$-dHD77rSsD7| zPvKi$H*R5nMz8O}!rvFz#alJ_dXV8zpPAHoGXc5k#O6A9Qu798@8Bsx>pH~oPv?Hc z(bTEy1!B&e1GjE}DY!3Fnk>t@a*4Us#^f*ZXZyz8ChdNa^Fe(Y)c&z6+rW~9!{F(b z_VUJ3@~;~wd~kG^u{S;i%F5m^)xX?z{?r>;o$V|m*^JmW@m0Kb30(c;#&J_>Z_S!% zRh`<}G-&n)a3wo2i9?77KU*}03!#Ce{875#weW^Y^#zYb0AZ)&DT==U>;%=GxdRJsHX8YsOd2pnP>5 zQ!PAemJi&H(>)iqkutor_AyCNhLldRkior^1+%4#8%EJiGhKq1^O955sBiM8H}yPD zEhiCM8CyUA@SA||p-@uVTxd3Lk0oi#o#XD7Lf=mc`|vxQmL^!_+os`kSL(;0=m5Q0 zv)(qHmCNNr0d_1`2pxXx+C^dAlxR^ev#;`AjCHkV9=^A65uqc8-0)ur5v@-UuAtt2v&3d1J+_Q3K2__` zPPz1b(~Sn>7K4y2{ZT;m+9)Kz^qjiBVcdUSP;{ZK&M48LdIN9~gIHp{cA!!7r&onG z(z(=+-;u_Ri^NHeTFhA0CpbBp6GsMYRt3-XPR0A#D-l&UN#lDFoY^}(p zVnDGJRAY+7MaPMU-!l?vD4pLFUeCFK3MP?XN38{k@i>t*oWXreb*0y|EKN1C9*rOJ zwADbjOE=N!7rs4I0$u;*3!td)L(o=}OC@w&hIke*N&HB3!F7S;l(itL{qK3hj}E4w zwL8{O)<>(?3v60h^J*wx``dW8pTPJ~(ku^pbiS~S+N-WNn;`dac_jC3>+db#PrI1q z48eMZOL|6yS_#ObIT|UgwJln`4O|ipyqta4uj-3taMM}DP!#`wRXHE-b5fD|Yeu3< zm+Q<_IyI&2Yx&I%I*j+HG6ihzNj#(%oA`|IN3fapiJ2Jzn9-vPz!WQ0f za@p*bU+}iVL#L&(ZJcnN*lqp{uSkZ8_V9NR&zX$IZm9Mg2ZsgU+F*!qV?Y08V|82f z-&`t^`nX%u?iGuctXK4sHxDbAy$DK#c?#5h0wDkXzL$NB=yEs zVL;LQHk*KkVEl375WU)lr*0w%`2c2U&F>{0 zl^LhiFWf{|z6Kev&vpq7c-zqpJ%Ms}+5uNzS5L1y#j_rd@3tvs;u*%e%`JK)bS{(J zfG25prcB%mSocI?U_Zp`#)fz(FH$+KChi~~@ylk{G+Bhi8h+sxcPf%P!^*dv!??gE z)=t?w{sbWv(UBzWOxV>>DrQu0YXI3h2L__!Wj(60+#R2}ro;nnuL!*t&+yA$WeEt8 zg_$!GF;5D~9etMsFU>Cbjtif^fQW3x6dVXT??3(<+BbTVL=m4d<8>1z^$2niIW`R& ztp_mM-XlZQ2gzxaJi`>1Vum#M_``+2jm?)k1DMs!OKZFk;7S-8y@>j{x0mJ<7^C!h z{%7;b!L<{L(cq4sxIOQt{!oPfv;I7yL%R`EB_VpHy0-obrn0ixTjOP2N_caYn=2)K zvL6dt%K;BO6&)K5!^&EK>&_{2Ak5*2v6U#+fuvc);_s86I{DQe95wX&mkv{_8cS7t z0aHa#kDCbE@{v26ES#UkPc#IU$&+3{u~E1IXUSI|>^2R4FRHRLzhr4ia5wFTXB1a2 z#O{2ug-FQFRrPFviluR_E0#|J(j&@WnqlzdLMjrLKPD38w$b_kBKziCTd zIQ}{rXC@DJRWiXi^&&z)&Ou^Ydh;Oeh&n1wuX*$&vGFH+jaH zf}%o_e5GC|imKP;@Fm8C9^^1z#xMF=p3*GVetMJwQAMzHz+4xH0ut-uu%Hwl{!sg_ z4fLmQEal&NW84O+D6Tb+2&6_~gKyG37A=YuO1{dI6L#7K0i5^y0vESgcfbmv__V0*wX< z*-D(`!1&Yr;$C~3141929~Ux8Wc`80=iB5*Zm~&ZtwzX2IV`SQt;{N$bK*Pr_btqg z8d1BjcsP3O$xK!MMW$!=qKaF7a)6$e6+Hjt_Km9{Qnmj+xFY+r@3 z#mH2tgJ$pTlDFI2peq#Jy*8b?^vkx?eg(aw*CO&kVTq#G?y-64?b(e@0WVZLFSA|- zFpq8;Ak3vw>}mUuNcmzm$Le+s9le+gfWn~Qy1Fn60%k&oWJr@O1yi?A{v%Z-QGdKxuY{C1mVRA@m!n=N6ZHtR)7YEjWq zqU>B20J-}=15O~q3lCH-RkXSA2#M0_$5`Rz)BI{_pJ3=3k47MR%pufjTdat`h6qGF zJvSF|R8_~By=;j8GNPu!4!F)VYtc)-FG#4m5^`P$T6DZL8J|sc7zM+;DiS>|KU{nT zk?wFvF8|!TQgx&E|6@Cq+9)soqSssEb13@3VEo{(3|gCv?_o-Aen)ELQ`l`wEk!X4 zC?qlca=a+_MA8_q>8^?bpUG&e_w}F5*;=&fi;(F@pYcW3nBN{uMWolI>%8As{g{BF z`Lx8bc|2|a+X_K+_XtExjLLraJ&r^&YgxJtkL&gm!MZbwd%j&?yCUyqpR|1MtC7cHK*{@ct=JaEFWP zBwhKvP$pZ#Uf_5`&&d{qPU-@x>eX^>_ltZSeL}w(RdL`c8FbY9NH9LG3=@Uu_ye`4 zq@Ih&krbWn1nO3Wc6ap{Eh4qK=y7u_t52W+zN@hs1-ecdGC#tWi!hk)sI+`4!+w<7=@leVB3TJj{eNY=s*7fB9JiU6TNV|M2vzqJ_Kb$H?xdV=~SB zfsJ$9RTPvBpVW{AGDEs|5CBOZw~8AYImvAB-Q!q+sKdO_zrG$Kdvgs{koQb`nnViuJpePQ&LO(rX*~jzK!yH@zqDy$|}WN&O4CVY&C{ zOw1RFJGp}&b-b4@b-N+dxo0+gE1d`nTuRdKunxCPIm!Vu&ljffs5&z#W}8qmQ>+}{ z=T6)4@q=Dsl#|H0)9+5ljn)4cN!Il)y^%(XW8|-m757G6eyTB_2DMrM6Wb;FHFCaU zB+i0Czu!huKkSeBVhCCgsxR&JGW_|mW1aZ+#?)BU#mg%T+a?Exgb!kt9%UZ@dv59- zo;omjk|^o)X#kp`(C9Iwu(qR{wJIb=^Z#s@xdlLV z!XG@D9`+TtxGcGmRg*H)DzP5=nzV9=)?I>80UJVGbD1(AbJKNJlkxT@)(0qs2M zjGzGjF+a7DW0sWD`<3zBqZ}z;^!!H8%N%xTecu-g9pO)1X%}zZHC~k@hAqJ#1@%X| z+9z}NnbqEpUpL22|M>Cg3(25RWs`W%X?)X({CId*9`XJL*;$(o&27!g5A!wYO1n|& z!UZrn)?lc*l|ImYgF*O+Z-(nj?`0G0U^PGw)soFGx~`Zfe6n=8oUx;xu5TnWMgCDV zAw=5!3$(tnGgJm49dOgRa(rmjlgPUX%>=VlyDr+eUVb%U+m{{#Ms=CuGLu@?zuvsw zb9?`K>N0!elBLi*!61Y+(sDhOy)u5`S4d52aAyo`oGbLzew#?4npVURn5sXAsoHaK zlmAkbdJ+1X@q;*{i)vuFRpQ`>A_mO`N8XtBrHP+}onMZ)slpZn5XRY4Q+uj%-N$n?KTDeA< z^4HYFrGY(YiAiuz={={-=#@XRLuon)2yy`3%bW}*t=Zf@eHmkPj#TG-J^e22NdiUd z$A>j-A!3M;U%t3pU}5~MS^;+VtA%z$wz#Gf@I7~9Q72t44jZYlP8cAnSYF}A$vC8- z&spV3>IDOhMMa?ReRD_iiP6IN5zyo@zuON zumyQ;iM3{Q#>kK%7LOl}K$Dz)ta1Ix7I!Wm`7m*{Aha15lW8OWhM7k_NCGAHtW~D zBpfVXtM5R@6F;{6J+}<^`CFieg5>69`r0MQ><7%|eyE4#GwaDFbGZ3Tp9-J!NWP%#@@1 zHAk_G3iqeCGF1|fN2)+z7zBDyV|_ti2OKarrM6+CTiohpO3>ZRj)8QJCrkuGU(bCW z5%OX@mOXw|p6(61S@Ftjkr-Gn4eAeE0hPa{$==Z@z_@K!&06#nxVzGG^+%b z9n}?ayq7`OKLn7_J=5nLYoaw#*y~d{5v^;FwmojdUTQ?}x26bF8{T;R3ZV7Frw8q4 z2F~7jj|U#rNCpjW!`m52xs1&7|8(zkXYq%Ijj|55@$^jG-ghr&0;3;LOrpB&2jC~QY5-Gq$H}=Ig^)> z*LzFRZeGiVi=`NQIT_xJ9lokEadd{+U=*5dFh7!-AgwZ8Z$y3&tp`Y=!lrdNF2DXf z^Y;s%{=M0SPX_bGPQ0I|a1^=Y7aD;1rswvztBx)n-4(q7!QfqDZ74`v=>f` z%#Fok`)lDDzy7xgzKUtV(it`!byX&IZ^2=YWdS=iwY2T9cnt-A|Ib8; zCzii{q;ciVpSAfDq6$wBNk-KE+JA2}#=E2KuS7b>npSIbdMLG}^ep58y~_J7@mpMx z9t2X0Gf-u4CUIDQzzgad@?v~u6OLrxL%*dq_|GYV>)I`BnMKqV@L3PKGRAvg?2=SD zCf`ZYKX%Q_KIa=;^ls6kU~A9$ncu(0??=_sVwlGkp{B`A*`8&#bo!Z&BK6F5^IefA z%a4t9v0q6Jz=qp#tu&6-kG@9uL3FkW|7bdC&E0VXMJ5b%{Wt4Z>fFKL%;qJ9=sq37 z;#>ViZ<}*^Gl-7|D5HS@Ayoxwf;FN;*EBMZe#ux~%j0fgM?k^qSMhpV?Bl z02|S)Urdw|)yPw+T5XU85Q^v63I$9dBTRnOjXF<-N$Rh5J%rg+~E{U)bplL)S*~ zi(aiezlni89Up$XZ%g5WrSx;Kj%+|@&O&Zk!r`nG-Ap#L!a~g`TITFiPvj5ZIT}PR z**XRlL~ftaJy?VN14{Iy-pdD6`!2jeWMfb0Q=ozIebwZN#x1I61Z@96aLYg zOJV=vYj>_wtAeBM$#4x#8oo?;JRKJ9_sxtND5J|H-x{12` zRMl?Y$6DmF9c*{LxoZ)6!i0tcH%-6=Zd zU6*M1qfCv!l`19F^o&XX*mCdlOZ4#w*#RQ)@1WG8lsAiROlJ~o`TN4|>Jw1Joh0iB zVYy;T024|h$KzL6|IozByQQt|(S!6OVV^_6KhP4hm7fgDQP6^p+w5TO1|s8*ygJ)u zvLTJJL&SQp*ZvP>P#ft|YVcM?DvN+TLB==wJM^QM$LW{X7{xOF9@uf0a1t zHOegAN?W5fulSfJ|3VM?g;rA9h3%8~F4tdUa%|o!BHJug_BPw;Hw!(+$)q&86DveZ z_1}n-f}-k7pKJUY1}!L7UwG%Hqnp{1);abs2dN`Sq%?AcUOOCmj42Q=*LHLoEs4Eh z8R45tZ1`O%AG-3Ic%F|i-3p+jFSUguf@bs&nNN2n>_f291fiUR+6<&l{Ftl3GpuFv zuP3r&s2aBPfSw&AamoJqaoyX(>K~-`4$eCa$~(0?X(H2U&?EX@R7}_u4hF58M8P8# zV@cb0ha-ua{h9MMJw&NB9}uj+mM@Fwd=3~2In^k|-Nn8$iUGQfqN%}b#z~u_JW8_o zmn~`}P>%qy&2j$Z1^9b6OO)=IXm5RO8O>Bp8rYum?S*qW^ku(~uAp6cif{M1dS zMBAh4*t)?tzDBKgUlE1m8tHsqi3}3*)vfC)Get`We7NnGu)3I0o@r1n;4IH$V;+x- zlz)GQq0=5ab*Tmro}3gF1m~`zKK<HzT?=O=&%YEw4mjZH`79DzVwxx&**q@7!QHkN;%;y zH7vNe6@2YqPIhJ$^(nA%?wT^3>BV>})&5-5-1}SIFps|Nc*M$1FJYUj-@-mTG1@u8 zAB(VuHv zrI4s-&cr&o2@3DKX~YQ?XJoRZ`?-b85Fgin;JyyX<{^tv_-@T zT-S{kSAwB7cKp}<_;H*-o97RUlRZbfA#KqwSp}xl_mj??_%4mKTU@FFJDPvNTtElk zfJ*l8f2uo3K>WrYYKzilW3!iNjl7RbF?pXuco=kU7_EntZPvy>mr8faEzYLO0Z>FtuK;>-I8o^DOy}%_rZ?e_@b~=zQxZE>vsZ(yij6>!Jf5{ z(}pG8=cGjpgYK4n22oN3^if^BoZDs3rkO6O+kT{rxh>&;xp2RY5;&ACOwdb8rzb7U z{QD`qGn4YNP@RiOhVTp>xf;NphM$>Xqai6Gs~AVpLZ`-Q_BWllNNeLc;PZtA8s6_{ zfJ|lD9_9{m_`@j1#YRl|7NA;-<2qCiDZCiv_~;aZ!~U&~V@GcB&M=Idh+a6&nTDBA z?`u=XpX)R}FO`(Y&+{(^!YAGuR~c`tc)l9#%<|HTP$zOsoVuFQURc>Vpfq0ah-rK4 zAS6Jn2P3p)`&{w`{O9x=(z~%5!W@V@10>GzyYPP#EoNg+Brxcfm{Rp|g|IM^B-eE! z!mG)z+dckmHS5!mVscgUm=Wk(fgvI0C#j^rw%)4l=;_&SH~eT}$j2k)*IDUa?jp%i zHOuy>C^=4&ogsu~441&`G{J+1B7Zz%{G^K?oe8+^?BTAIB(BWu>eyv)8=dIZ>49M~ zR6~pJ+M|tww(d{!xY>ogy8L4%2VZ@^B{VMO1CW)NSj0=y3wuSpEk}zJfjqf^T6_tc z&OCJj+6D}(n_A2lfp6(g^){cX*1Cc-)xrcEspei%?!)Q>fkiMMEAg||EQM+i_=?b0 z(ZNp2vgFudy*EH2{h6%{Y`N~_f(sXtgUDG3&i!5eZhoSNL{v{Z%F58vipWyHqK_l! zjJ&Z16vT9Dp_}o|M`0XUuTqp(`IBoxhAxI#ZeWi9>adF zTW+}@B=D@UAipg-iM&_Z6A_kl&l5WrK-P|Fyw1YA_#3j{Qtzvk+At*VFT(C9ebB!6o{o&L1q6fd~Wp%^f6duRvkj&_abjqn~ODm z0zrTC-zn#e605-#5xy;H-FBRea2~$8Kr9K8l+^l0}lZ zCm-OM^|a#WmO;ZilIIj{=gZp|Tte8`vI87qr!sM5#1Ue^!RuS=mQKB=JWGGRy1Y)qrY=QQ8Lp)OYq%Eu*oz^S194*K^ z*!*T}IRzNu>L3T;sZwCY0`JvkeUv?LibxVK(>SKWp0%YZSzmxZ;}mbO$*Wk0s5~Wa z6hF&k5xpc;6BccZ-nL=K!XE8t{TA`?bxlL$Uoe}QuhOxm{$Bw}ENPm2SDkwCzkA;4 zPwErx=UdnQ_Tc`*wz*PX8RC)NQ+qoH?-EXX(>SVDxWtisDdtE#V&y{7JcARWTB@Vt-8i_4Co#S8^})SC#6{ zMn896@y$K+HnV-`k07Ub3%+wFaEG~7HGS`p6&7~TaC{2I9S4w++A;QWf|2r$7|237 z_a+a|XU+i!+$nrA8oTBv_6Qe06Hl+@*F!~k$MRcpY67go(C`1dp|Ahrv4;Osfxb3_ z*k&d*6Xwp4gLRKjPyX}=9}t2&;2)f+lK;t6o2w)zXpR0KB|IFHz^A7-9xEUM_xRTG z8+YN2Z_5t8PBl;kooUyfoCsH78DqVIs|)`Q4b?h*jEqX>`H1&qG|5DefrBZ_wGt!m zG4A#|7&y{&NHMQ4nfI(qY^z!G3zd493c*&^o^<^GFeA<@#{YMswT-(DERf?1EbJRA zJrXjd_Xt0q(*DGkGh7-piY)6d<7Hm~pUked3?96{xW%+{P>t$we|+*^lt+Mv4beIe znj#lM{n8`oLrm$t+}UQ&d*kdVmA^8t3;#i4o0_%X!xya6x?l}qHAiMIn+GIY$thdQ zRX)hP-guxNhM?TTa;3%S8Oz%5(K~rBB@nS|pPuL^#|6kgOwL4&QhzgfW3rOkE;~L{rmhC@X_qi4Ycb2Opm`^y*_@of-7)O(#8D6+qbtTeyEgkuHTck@zVxv z#XlWBj7-mqe|?h|xbm2{F1P1EbuNcarbJe#(=Wl93@qn>t5bpVy7y<@-!FP6o~Lo@ zw(E;n@-9xjaad531*oxJA}!|nrfao6-&8-y^?4`?R4^@{IQxqJpIXi}f;BF{G%hS) zmTPiz<$L{Qsx7mk{)v^(kec;rd+3_|r!@K;`$JNNSeh)faJ{P%zN?$_D$_5H5>(|%6( zx8h#^`}L-Mvg?+9*}irCk){K;*_hr7*6#f5{#LuT_OomD;;5;0@9+Q0kiHQX%gx)k zotNdFYW}`mSJcaAKRok(xBS=Z*)}!RH}BRiVY_@~oBywHfr`&V=ZK>}^~Xb3&o|iQ R*9R6@gone- zEn(SkhwV37Z`E)|06+!+um{M<0Az4^0>Hi&cP<04{}1tBvX`;PL4phppMZl|dmQ)w z$p0fG{73vRCEzyu1YBNgkK?JWwXenHcR>I^f(#C166|sCKY9W#)#A2}P9nfD`aP z;w&!R&EDP3H8Dd)B1|yz(psS(dVK8Giq4m&i#rpE?q`4txglMC`{nbc%#CXx601?C{%$YnT?RJOHtBPUQ$~-A^*QyO zccnz|Vd>1y{$i{3@BHP7`%?orQMrw}zt@6amY2&%qb!vbt8|u{1lb=5O@*S~Do!gm z6<+CV4ut4wiaZW@@GgKftwm$(%fQ0RALZg|q&x?d4O-10`GN_Kk&wpJgq~f!*jEey zAl|QguqOALE+nyZSF0XB~x8Kzd5>GcYy9O+j4}rk~L3@ zd$UDF5@NO$6(p7sH+_qxb>8*#r{BqJZwnQQu!6~CMI%9sJQc>LjA(9%(ZP`re?;)-KMuqi{ut-{1zOGiSiwftkAqM z8EN26mlqV;YJKGb%hvLX`Qr3`ub9M&`kLqw45j3v1d$mCxS+Iv|5$l*pcGL2)Oa_$%N#zzw)kNhdoAMB2D3a2FxTYY0jo$_|cFxdfo@Z*2LZzGUu!_*v z8P!R4Pk!)Cb;nS2v0ik`J$JRDcN&p$CDDxo4^SBJQJ9w zL!NvLwG$gvn&zfMmmR=0#hG;O=At%z`|692O}b>up$ooYv5JYvBupM=GYL!AGLF%6 zVxvw|l#>;GhUZ>tZnk9gy!`;$@cFd$%Q00A(9MT5EeF|fGmpe7OJIH+B%EC^KkK^W z`y~ha&i4z0QyT7a)Kjnjuzuzx!2Jr17%s)LRoFj2EM>w-f)|nmECNIFHLnHwE&s9n z?(BaX2vlU96Mn)*7npuXUoa-~0Fj#@%Xpukd5+nL!B`Gjp1Co9w_WEIYb>FCr+t+9 z?yuWf-63!JK2}rU3*_AT;Keci`Th%!!kVe3#I!yQ;u;FjF3&0jx4BroaD|||Z@O)L z_J!v(YdMiky&H))bT<=emjTT}|vgcOFRwdCB{Xq?CJcF@$b*vpWX?)tyuN%~{= zC~)UJ$afdLrDs%)l&NMBeDYe9p~blK9Xa~z zPheJnmTe{wywx!L8ie{LQp@Yk5H^|nZ5f(=FO`<;3M^tLO}q9|L6L$)NgzBK3pgP04!#dxx#`D5%|FUooS5oB;hVA z$eeyffQ*n2N3)hCjX;kmEq(E8Yl9&8sp?s?%7qyDN${T$9Xur?Bn84^?u*a?;cKC& z0?-T7a0r5xs-Eh;wv~`E8OZ!&+7BEv0sb?UIb9GLJrKxX^|URrFdXuR9VC}T#%Cq;gaSWDGcK8mF_{oJ^oXxS8S;i1Bp1_cDSSQr zO^nKqKK%s_#R!s%noIaZMMi=L+ep}ZL`G-_Ly4w+dxTG@_Fn<={|nqgWQ4ef8%=+V z!*Ib+N$K=j_rsoRzPvu!K0u~13_)*pDv3KM1r9srAwo~oq@f`}7uGAwcx#6{3~rc& zH<&%*&pKjHG=!n-lz6^ivPxgU!&MCu*mLQZS0`Rl&1dtAQ0%q!YSe9{8I|90oDd^z zezGEElC$BJw;IFdl2 zKDih*)4xTe0(tZK80bp~ZaW%$bjV693?l4&G7+(#U$J_o&H%p%QTNLQDTforb0NaRsgH!ZJ z4{Y=Gd9d~FN!`T_{ber`U>M$%$StH8iqk70NDg*&okvOD3xJ*QT5;gi8H&0}PN3vu zq`e2+(C?53k-hwn#D&vq6$~{^t49fh1C^FsNYaU5&+k2OmsrD(L_?lYNu^r6__2fbk#5Scw5?BofP#QiX|%L8|IwQM-*ziN~qG6@3w z!F-oWhG0h_22hro)ye~*G-CY0Jg-0DC<=hGS%cIAfi#2F5|b4ae*i$9Dy`Yp zMj?T2?^65`7^;VVw4(BT%nCnk1sN!--x>(kr5`k8G^e4GN-1DI6Sp8vCxMsV0~Y;$ zo79^B^#1EScP^!+xAe4^GKC#%5om7!Oy8tT9;}%xtRhTZ0R&}zoX=oUiOzD^5H1-0 z>tW?iptPi_VdiRXztP;t{m#{LacSkKq|*q1ucIn%H?(j?Qt7H0aB%sFkVD+4M2Ssl zN$l;b^Ung`4rjis;!l(k2mLTFA~68KMmJeJlqMPoE7{gPdYcSE3hY+y)KpT@6-?^N zd;L$ED<1pnSaw2QnKH8C0eL>W>-?bUz|>s$XqrZUQfNw-o+Ju}()S1~?_d z^fe4TQgQ;%@ z_d7Kuky)xO@$KhmnCpf^I6$lphmw_hI>LS*6xSn%4U`Yl@~EcpfZgd!2an+l=igD; zEqow3W~=*Q@(Zck~eEA?l@yLMF8whVB6FLIxsi?WLH}?n_Hrl z=0xD($9=8Tk6Ft`wQ~B>v^exd_qz81BiJfX2~O#^cKlUqn0cOCa5$C#HggS zMD>@-!y()&N!G3}2)4#t1c1d5p?g50zgQjTnoMH!DoE$Khv(DR4)(KcOIi=yK}^a# zb<@0@ex{o5t|Hg8Q~jR9;zaLy7+ya)>LHS*+C(JqTjh{?S^FyUYrx!c1b*%Jr|F&l28r>xl^|ca zB>}nQA5oC3j@y)M$H}EH-neP$ZW4Q_H1y1er;3(D zO>%ItzS%H;xbJt0ScCMvH0>2eG8i4z)OSq^RCCLHu*vwIpDM&EH~I_z3^i z1H>y2!1?Mph3^5#?yv+7d6cSq+-OIIX21oyvDUe&{|CYc zh+plASOOaSWCyvZ*pf;!Y?!6iS9gR4uUFwsKtAjoUq&#dq z_Yt;NF*wlsOI{BzTwjn!n+#oHSuA-kevQe`BA-&;Hjz2>_COOfXLWOO-N~Z3$sl;p z6I1YM@hfVe9yMlJ?)IPv4Y9)45tBmxYeh2o-c$D^Z_W`!P@rBeCCi5k2$97RyR;aW zTn-SIVadY7RiV-7@_@vt_?&NqlI?F%i^hThKA>}qHa#Pu{%6I-kn63Y+_}Zuef`d+ zC>Dmaz)Z4`*xbBket3iiX`4&V?tB-?-NT`Jr20!KCI>`tw$sD+u0*B~+J0+{zdzLd z%Khq}B$sz)>%DWXTH@=^M!KSq>Tyb@6%2b7wCQ1Ufls8) z`a+4(q9q8*kBW=U7Yj{M?U@I>3wM~wwV1^(sITUhnGdmXQQ$ARcJ42!YJc19)v(&r z%XIw{bav}l{)Vo?m$-MCmS~V0m*sPs_uXd6;L~CnYMnswSJ|`Ok_eXXOnPul1!2do zkPdG)d(3+{_PDl-?OY^@W&Jd7yw(4nF}EtHa0)ATw_Np37w5pJyvr5~Tlj)<1-NEk ztRAob*sn^l%bi$NC_(hf$Zd~HJ`4Zte04%93s!+d#!$84#UrNDH1tB`Xcm9!cbXb=xkc+-(6*jD$caP zRl|oqfWd2ackX;)&^ZBiZ|Ck@WswVed*Lr_L53AUUJihEuVzPlneQL_x;)chaiJ5k zci*+*$XIwN%gLvb-}0E(D4{MQaMO-D0M}~kt(z(|=1}&i3OC8ZmYauyrBTF_v65Uu z{%bdUQc(SM&!*tPwr!OdVYvQjim6FrhQU*dMTqsnQZeV$_H&|E0@JWML3{}*tei#j zUL||W_15>fj{!_X$RrDlAwC5GM8MP`T0VG9n#J9({(LojY4f{}2&RTC?o#$^W)mFD zsE-wcwXlnwDBw*`H>+~Yq(`(${Zy}@jH){+{E6$-SB-x=CO;8 zTJ5PEc0!D%zoCog9rlvvU?-B3mMux(EO3Nou;UKtaH%KCas@$xu4V}7o0~A*U z#6m4)TS*rkX!_6{3A@g8fBs@Bv63^}vGT=24jlQ&UI?zI2zY&-8Z+xcmcxzMj5jBM zT*oP{+bwiHN&DxWH&B1UTSqEia5;{@ZctNWIQtOG{d&uTuqtAZ>e;$ zV8`f{*LC?N$nS?qH|-?%{McYC=OiA;;O}Nh2bGSV@Qn9}R?@aoX8-b{%7?M(XGYb> z|9*kbRlkr8g3#y1dsRmFqy{YYtzQ{=DnUqpb9)d#jqdwtf!P>~H*HfFz`_Nuokk=% z;d=rTbn{AuyIxdgPtNG>!RJ*i8i*+=5*)rdzI=hw>lst`l_6a81D)Sq?NJ_B zLZma5!{=0iW(@y0uRIcisft3VOy%CJDJMr!w}Vfw*~|LfAT*fKQqs1 z>sX*45t;?FWYlew?XbVjZ+~f(Z7}27wA#D#gq}gFz+_5uH&mb?1Nxl%snAv_9}swg z&EWg}a`EB&nE@U#k~l?7yO#?FFt<&*vvT`b@3D!F=MJae29QUbgeO|ucH-W9Yrem- zy{rw_f8L)TlUpu@BE`Igg9)4wnLiFxy2-eCMk@*?>dvUUKKO>C=#jzmiex)y(4ppZ zE7PMGlB}u6yJE5Fe}7}LO^K}$m?2S5M$Xy3N@Lg~>=(`HSN}8BrM!grY{u&J2N}3Uo7F4fC6%QsCt82zgQYa^t&hzjHo3JzTj&~16$z0EBAZO&kgmus1^H$~aaQe^4 zt@zs=Wg&~S95123jwY5Z>peu|v9rj87yFxASCsn zEu#-KnIe}H!y5E7=LNJq^8fvmqe26G&!GxB(agI!oU)%R2$mX|6^ianleH4$=)cuvI}cB@eBZyHbVZh85r@&Q(cn1s1M*?D|aJWZbQ;o%?V{=fPQ z+tHh5l2X(#KJ14ouZ_3LHAKIR7T&&j3{qD?obuCG(Wq;q7O z`H8f~kWNvcv(dcjzf~UEEJa~>R#zF2Stw~@`QV6P{WrXDZ&2ZhwW&xQ)o3sL)`1NQ zB=@@yzlwI$d$nVh_^>L*JD2W}!yKq=@OjtKF0mW^^j-tV^-B4+AT-TD@SKuM9 zdT!_MZddasj_>*1hl$!IX%GB162h*p^{UAHH3SugN!#HZuFcgWHdqqhi1pRRqVT0w z%Q&RSy@v0cXY3ui@GYch<>`3L#*pN-PaEaamTaMRG2gja!D*?!8=`DOVc#da;Sh0> z+FP72Z^M^hdW4IyB~tGXq$0>v|19nS_DkIvIqB8~mHO*C3UqZSOUK2=c-?_zb%5wa z!#+)nprCu4UG>478|cQY17fA@ZnJ+!YLwH`_OkYeHi}-zZ{etlrOyzctjm(V^& zErlPM;>I^8m_1|#)rXdUF{D{rZZ;N*2FQ<8n;QfAevFno6!weJ=W`xrArK7T;vU|i zBbtl*ySmKZayy|FH|M)-@s8!ZV8`q3*t}aN`-(!W%<}5LSEO0brV7+b04@?QZVj%6 zM2vm&Z@4$2*VYx+-Dc*rKfv)g>)+|{%*PZSY;_7pXA`Qey3~QtY_+KII|cU=OBI`K zgZxbOTZ710EmpOc@}%M<_5?l@R%45*y^JnK?Eyw$Ar9`}Sr`3M1qWiba-I)d5eSt- z0cI_HLxu!Hf!1aeadF%XSO+`^mKjo`9XvDk&X1Z63xwNrQ*9~A4!vb}sBJrOeZhkfU*p z;OiN)>p7QOwhOdsX6Kl&Pa3$v7VyQkNU?8u$zl66&ut8{GSqjoO`vmRWVWfgQ+1SD zXoDG_)R-F}egcnZ{6;PN84X^Ki(J8nlJ?uN*#h`cQQ{7TMO7*0Tp zu6rkRv~eYPchnPn_r|gFSkzRFfc;{!n$cI`YWD}E+B?Q+PM*5d|FO}RhlI0|EN6_+ z_@tR_Ew~M5y>9kqS3sYFu*ZaXrR%Gp8DmpW&Uv|^{M@{TGdP1@v7IU z*^!O^Z=W7m^WUu5_d-vC^aAPoP##P1bf)k}I6;by-%6kMcT{OTB0_l;J&sCJIUF0+ z6Fv7{DiM4XO8_wg7X(%`7=9>}bFG9-sLvFvp41weo!wDlpYjyej+qe75^1O)G7Na& zz9vQv8JE8!-UdlUB#}=57u`V%L*7gGp_+VuEsQ71R_RehA8y7$%SC5?Mll~-HGNOQ z)0~Bjx?jX@n(fFhfcaFM+&h+04zs45ehZRj<-6-L?sMR#2-Y>)>K*Evw3!YW}r6}*`#>L)n~ zqrntFnW1aepSTyn-Ag{Gmffwjsg`9zYnGZU%?-Y?C*pOn^rC5HLG+(d6SP?rD4lUO zxQ!MpdeH{ZkeUQ8WR^tn+jd;J>^;jTzut+q>b%u6CaS4PO4#nu1{UvX7hkG>>2 z^%Cqk;v&L7zaEyP54-rl=GOsM93$U~Mh>V#_n25s!sbE25e6W3A1}hreYLles;SXJ z8^yX)beBJh@FVO-bfs19ZkXGRpuO zVM`h>+OGzGtzIQ?F7;$odh7{I$?-7vheIBva#4#Lb3D%FO%eU+?%4H4i1X5itoF9} zR76h-!10c9>~pOgR5XVsPdKrL({o1UR|x52id+MpnWpQuo1Ol(uTEJsaD=xD{QGrt zhHiL(QQF>pSoyTl-K&_#9unyaV(DFtyzssSvSe@&FAVQsxC^3iw_ll#ow1pH!K)QR zzMCJ?o*D^p`1PZ>#^m+8<3GjgAboN}sKhm|u?HU8Rt8u=4Ev&z#=p-xO?98!m3gNy z65XnJcH8Xu1TYI;dz?asS~tk#AruuImr8-Kl+J7gu`44G>TwGK3OwHO#syE83i$pgHLJ-QO;el==% zP(Q=8B0k%Ffi8!1@RTEr3_ZVjZ?-h zPV$?PgOXA;J1z&qw(<_Kzl_e(s>s;S5jVrNNS?;5cGp6+9?2>8FH>NdY}7Vo18{6_ zk?@0}rXtA+XBx)0SsKm;eTwnd9bEjwqlh87AIvikp0E`CXV>*1gx#U|a5%bE2!I zZV-mYaz1^<+bl6rW3-vMmg*MAL}GSgtmFIqubX4kV1(l+hHmcQvFwL5lZTJe6i#Cq zS()&R4o~ti%>{epd&$yGPXvLvUMzKzVz23h)3$|eXQP(%`xoc@kEI_+E@O2;?HnaH z4_y4CJIgICz8c>1G5o?t`AV7dQbOkM0>77mqM#FW;<)OpP=G~7UFFtWF;LB!=1JD1 z(bT*nU+i9F9NMYmykom`r{QIwT#=NVTwh0!L2)LD|LLge$V-MfBlMyD1F}HQDVIeg zy5D;#AjEQv0=4Z)-U+s6P+@s!>^BVVgJkp`DeI`qVk5U}O_i98>Y3<6nw)LMbo;8- zndSB?$&kx}VawBbIql_vyi@)*lYZE@rziPeKX=z7ZB;rZ-ukcjRuEbv_&VL6Yg!(X ztAVYqXLbl^J(;4a>Jxz+<6hL{NhNKM-`~*(+UgX;ETz2te#X;8$-~@u@o%tiPfqe% zk=v9R`M>PrPTP8V)@3?7pE`uO-yus?pwD}bQwKD!Wsf4ei?fC}W7c3>G~f2dXQp1& zC6U*Lm7m2z2rjV&YX~{P?b9hvzXL*}AW7=0@c1&M3^rpH)F?oc&DGIf$=md|nBao1 zW#iS^x@&rsJk#Vwjusc8>*HAMzHd&g9)Y&}h9af9 z?ki;sd`Byx)8vKixw5+Z$Mv+Bc8L!N=)Sf-X4g4PolHYhMLRgUe$3uAEDx!yzpxWI+RO zM#`gQ#gKYDGi_6<1!O-&ta!@We~C0^9_TW{B6zC z`kZW(aA4F%#NBTVc$yEuQb_x`?SkA@qC}Uy>ZnJgNaonNZxeyTsuM0>FSHeoLY=?q zz{a~bu3vrqb(5RH^iD>8yhHhhDtUa zFQRqcNy8b4K#&(+#|gI9(oXQzdWBU;%|fhN>cAzmX(y9YbWV32;N!Jy%#qHh(Gz%d zv+^S9kwd7XUE4Kex_|ur^hCFZ^sG zq}Tw0KFy6r<}dp=8AC@E7NW5^rK<;{|Cf}X@(0d5Z`X1O_kfq5%nh^qJ6Al=gb!HOapD4? z;}{vMQL+8dx2Xs%&AJPB(P6Zzm z)a!sKE}bjWAH=UGq{7T_@rlpqFT~i4fG2Pk|NY39j&SXN@+>YG|1;ls@K?c*4Ag$o zvyaHQya`Ke6LGFfPHSBA`lp0#@r>v)_8O{xQ;(U0QmD$Z^e57M%Zj1$gOY3{r9=YP z#Q^KWWI&lzW>Zvo>*?W#X^v@aP*C>26N0bz(p+WO<5j^SJ?{l^6D3U2!0*YFNlP*b zG4XCN6MU@ThZua)HC}rdZ2fWvBSUUQ;V@VqO5*{IJTrqOwcDd*cs?b@o2cae; z)+(ilZr7rr6(K4gm+JXE-84tvCAUq4sIVxM=y-iW^VN0xVG4Z5lU$>OQhl_Z9rJ0; zNDP5LoRhnH&$Q?6H7g)X_q~-TURvOM|7H{xHmq1N z49|Ih*b!`y&3nh}y8Bd+U6|bdcN2QJly7YE&xO&vzv?e#_YySU8FBe#pZB+Mj*@5d zy4y_rH6C5^N7tn;KDTc{_GIS-3kB#}#rD^waulz_L+0j&V$Sg(4lGBhjPJOH=N0WI ziM`{(Et`~W0cA8Cw+{MgRAn2#YON7_f*hf220HESSGK-B!&8~%+k-Fm>F~Z**&s^F zkZ&QCoKtl5F^$K*YTyMo7qb4Kf7~~3{%p`t`C8l5sm_8vTuHr7B%@` zWeB~`-86k#mg7!V-=mZjOKHj>Ci4m!ZL@}#Qe~47iXl-pWuac{pB0v zdzO8r4QT$OW0D(Jn?djdBz<3lC%@aNu1hHGXfk6Z6zbHVm>#CQl>3BJlVM6P(kyR2m%UAZo`leK$tfqR~^PD?>Hkq;6 zE5@s(W?KtW(8tA6HCFPMbTvFs{JmdJMdq5=sod8dZSv;$?(go_hy865A6JHK76aKs zQV&1qrzXbrN9|qi?0pxsLC|nZ=Q@08IZu^4aQg$XI?J0^Gv_B|ky{tMZ#)cqoUV>BkiQ!^EqfQ9~*r8ZM#YsdU^#OSXY*(3Irc;AthA6QRvE^A&?TCqd0UoeU=Fgx^Pdu+X;u> zDCV~ME|1=&W1hsT!h)e|`l)X>(kT(!C2bw-Uc1(Uso5))P6Z3C>c16?i@v%gSVDG! zxGwpn64ii+g{5X~&f2Hq-n-7(G5;J+=D9yHl~o|$!JAUmyO%Pl{6vt!+D2B~*m|oB zlsXhJS9r}waDL3{d@X%UB8lmJ!910%b0zk;q4g!dC~M`KUbf@a{D?6&gK@(;uJ~?d zYh11oSPJ**rdNwXulEq&vJ9^NVZRyf#NR&%t8WTld_}c?r)Byv{qf;nHbCyXi}V|6 zj%)z)mBe7R1xFxXwzw6HqWe`+}Qrw?Q*$dMyg`4*QTIaEPd{S!@qk{>ES z)s1`%BEoa~Rcm%0`c+~ueyt=uZ=>>=ebt!1mHBpL=QxYftZb;uP>#RPx=oQa`hukF zgnuS_LhY|0J$R{wsqA*E@>+(nt2eDPz`a6R9-1N+mHk*~MR1E%m;=hkk{Fd8ZuP0< zK2$~BHe#s)y8!{|#Gtt{xwaHXUpjev@0EMRAD6f(-jff9oPA{pJ5wm6v9M+2@5{Hq z^9@E}$(}t%nBP24=xgZu#KADd2JnsNuOX5wS4XBBq+vmYDpKb02|JbEOyi7Y3ywkX zM-i!#Q%94rP0h&b!XF_?o`GI!OW^_|lp?Qacxeg1m@OGmG~FYuc-=7}d&$ zK6q^I9j$5}PFgw&l1sQv0por0Ur~f-c8S(~$gxqsmY=9-j}pP~%E6*uX?{kpKd{eR z|5c5OElIXJ+`k?chUVPmdI8P6 zr6O|z@L<@3^ei2^Lf}Vph>21ibn&E1^U>Y%+?6syA>z?J;6O?=*dFh0Xv0T zYmwjGNv9hTduGQ7hZ9GLwv_y*WeL}AFw}vqd#vaMVGGyw&_rba?p*wY(+5kCWx%sW z;+q`;NY~yjqk??Xn!Wg*1Q^&-A0RL#xE{YT!|!|7w)NnCR-?JM$Owv&`I=y&qO6Gnp%6bh~Y zJbb&w9=&IiH-2Z|zNJ%#Z7`KNWLrenGx67vjkHT=++6I^%of`bH%|$rl|deHPWgj| zo2~T{8O|64+V5YMHvQ;X@#yTvKfi6kWG`3J9V7J=rzQ|NZCo*R$7ONggdE$TdvUz*I74Icj)UG-Aah7MXMPe|`{Y*PEBiZ0ZSWI=&0wC6|4 zA_hsj_rR}bxM@}N#fy0G%YQcTVMGG)Nz5YV#*?9Se| z&0x-jU%7TLLBQ>qwT&)q3(}7IOgh8YxO1~1t(&&%_xmjS)Ooxa)eCus&(jHcf2biU zuu?e=^TAh;UVNWt<=)`WZ$%tbKvi~Yfq98o85(1`=3(FflkVNT0=lDW(^{1ORQQ=M ze`~iFlwErZoqkuwF3NAW5p3fx5lte=%{A$5M~HeXnNf4?EXC+X>yn$P!87w#(Z}z5 zeemWq=cR4p@2Ir{Q|G)&WjSD7*CItAVTOF>_l<@jAs2UZXibb+uPsy>z6PFwB)&(` zSCRAWYMF8oK`idc(wDwnMY0=0CLo>?48&*}a)pSG`!)HDbj%!b~Mu za~*|vb*lqWy`EeA(OYlXzmIvkAqNkv(<(mvu}V?~y|VifI*~{Sk$kmr@}cNgR?-S1 zzs871wxb}vByN!#TeYvMP19ZTA!PPK{URTVJyEpuC3Csq!!55LXBk`2z)MKijks0a zRTulQrqIN6Z;BjUO^KU4HgFn~_C{J&xqSIyC*oe>s6Zhrd)3PdlK!!;)vyt{+wz9T zRBF_?V`+7qmGgBsmqH0R{jVqDgn~P6XDI%HC8NV>i`vf_;F2qS<8?luko>)OfTZn1 z)N5Nd?vp(29Ml(lW%m{4e)bpPFJ!gi^(f2t_C+@Y>5Nl)vIK{~I|=pl3iM&G$l&D) z)gn_RX1lC$ol{eYwCkoeuez1^VEw}ysAaO2y0Pf+go)?t$FuBoSOC8*BE`*kG5yF$ zRRyvCB5?0Pv4vwq_6GBm!$*0GIHVJxKX@^nq8h4k?Vvg@|BtgCB{a3KLIc-VKK$5X zDZ{tdh`r|fwXbk@64Nwvcf6+_tr#a#b(w~mVB_Zo=tI**rAA;98>1pp?m3#L@24F_ zX2D%c%kh05e%3De=Ifgg7P%K99|Jr$qKQ#6>nB%PRAE>CWerAe*kEg&k4;At1@}8V z=IDP$A(dkr*K6aY_UARC1Z87ny`jz6p_maC;~?mjg4StOouWu%o8YedYWdaeMvglu z$4!Wd{37rtaw~IJq1s2?pQ4^R9Z<%V2<%oqv?i1dcJvqSDR~-5Qng#q^pkD!EVYGY z>x19J-4&9ZeDw)sGc0>7a=8c{cjX*xZHk;&Dg-YqE~!?VP!II0zTVEBUY1&~oGWof zvSZeMuL2Q4@>y9*APRDL%TulIx?IM1MB~x5+d4flt=KpogMNQ5=yX$JP9*xE@5VVE zW&xGGuA)U4eg2%&O!*l!6-n%1+_CtE!}OWbo3H+KIPd(9jyfhq^CObKT^2j8!;4rV zz{!4HWZe$20Ft9)%S;_<<^Ec>R9WKlsawYE^pWy&fLgj*B5bOLCG_V1T-ah_X5Atk zm{lrQ-4K-Xl<*YvfH^#+C14T_zVjUIXdAh}{$cq9q%)jC6DA>+^vLy-HzQstrCtCf=nZ>*-r`1Wy=*J3Uhw+oBQ z$X61yfUxoVQQ%GRrrvTYM#I`Y%ZU0;=8$VwfsMi~`1*)*YckfJJ{R}Fkj{q?zup>4(T6+@P{PQ4Sv9xcF`fb|;+HCFgVy2e^=Rqk)P^Xlj>WI>$n|CQhfjn!;B0UewIU%9 zt9V*?FB5^dFm#gM&q5D>U&m*Jm;FpK=-G)(hh5WEzpYzR-|XQXnayl9b2&_gc;15noPK?i_;?Cgga zAH$*KR1_}Qb3h)&qS?+BPbu~rlLBlq8{g6t@>{k-&|dPpz{+(_=cf6*<5G|t1}l{U zhox@&lu?pKmuipoF!iYJ4t=?EtN5&bFL>&qBx{(ON>1O1JE!=t7W2a5rW}kWwEOnG z%F(3459#bpCc=w!O1<>h8Hm2=z2Nu&ZFD1650<-{t>LUNE6)X_Eb`m+)H?KCbtzZnStTRZn@+yCvN-jvy?Yol|=$j4pp_5&If!a6H$Y088i{&5avPovwn0A979lG;ab+3?+N_idel1fwG#&T=iDv-$rSQ;)K zdhpLLRHHFL)envd&th*%7vy`H%8b3IDXKlgyb5gC3`!XU-g=Z;{t8gtQbA1q(wkq% zGt^fuQ0VIdFJCoPpzPO9n0-07{?Xv;ocSJ$Kh6|&*#39!)1i#;T>H*BY7$GCbd}aR zNB6_=eS_$Bc78HdKcl$_zVF5-abrqom4T>3dL+0hu?IP67Hx-bdBQOGVp2Kg2hm*o zRpu2_wuX;A?RJlK^X=Px$nk)>C?&p0;f+i;Hb`mk{b9K8|(Fe|U z^8CBmaTwhu?I&evwu5zQ{%pEj`CL0VCpgVFHM`7Qc4tUPS?TpEZ}7$7^M037eEk?< zXalj+`D3mK+!$8TA4WCTSmV6utE#i}Z9`p-JXDQ`RKb}Um7w8-`+~lap#y@K)VB4wQ zV#Sa4)ZT;dM0}TSmcCx`ITyyl`GvbgbhcwDT1|Fnz5LQQ0rxUv&hXq1&2$U`eh7OK zR*m8py>|oCIhX(J4};pDwrRVq0C{ejC}HpuV_|s9JASuON4JN4 zix5B@$B2fQ2YGGM(4Oj!EQ@2P+t`r!XUT*Rb-G~fy6LnTgWO+6?#|jVdN9-@lOI=R z@u)-{uVIMKQ<-mqkB(@ew*Wa04Vzm{Z}J!Rc09QZVK=%6iOaG{!eXv|P2a|!RnlIW zY({3XfF$&1wxT{2{MMWKBKh38QlvuDvC5tv-l0Z*0EUee{T02?dOwgrRM=t9@e{{^ zVj$b*CI544oFmdFPN30OY_?E0o^}){>mM0E{p(`0*BKWf*mg-hPbCRDtUnk1 z@InO z=|kTRjZ|~K$JkQUCp53xmV}Ed>h)pj4^2cW>xd(V=})-KXkLkZB$^waGJR&zGJx&J ziTBuDIcA(Mkc~GOT7x>4*-VzrC9NRQRC^OKG#bX5utHV8Q5poBWlvAnISz@t3?Lu zE9sy&faYrCEWK3Rv9H5DDSIJeT+3J$$@LAq>)wrr;@z+bKxaG9l{H9>7V)C-l}nLq z$A9b5WT67^rWz)T;xeRqAGoqYOSxhe0ARrT?|%XM`<6i?6TMk$(@W^ zfqx&T=%hIcZMZ??t19&ViAg3!Xpo_8af4u02V%~hcfp)l?~{L+dg_h6VT^6n#FDHD z;!}ub%|C`Z(@IT3W`9;&R@W|T+@5t*W9VS2@1m0F!=lj(6kCM)nm0>NGn9@jay)$I zO{^6#Oe2%z6qAQ_rmz}hAj`drOQ0F-Id|g7uY1DF^}*0|HFm>;^m9=j))8T8{2ir( zg9KXyn@xm>z}tXFlg5cjU$+H4?`ehvR;;*{1N%ms`Y5Uo8HTCtm}epPMVwAiU{)MmWpx zQ~S(d`A^#Kc1o+Vf|OL^wt^k$#>P_2Dvu>T1fH>>no;&G5vN8uOupk)W|BAJ)BFwy zy?8Ie6cR0LjIJ}KsryA@2H}jPu&63I!Sd2yNha+7>*ma(*=oB5ZV06aZK;}-qQlFTufJ`?;n-Dfd{qtAL>SM+;4PvdX`g%4WB|+JD`~zIPn1;v&L|CK z^7}%(6l85+%>7{UDWrak0ZLE!8MVlD4UMh9$yVokT??@M8QmC-&=J_~i=@i|dZulH z7?m7Hy=UH47i`s@Zt@M&fa|8-J(MD^vA0|?CQ(cfRES87sc5I6QQVTLYWF9|pN)x2 z+BKQ6%ivS0y^=wq=Goqgc6T{Ndofn79h{<65fXPgN^*0&M`7Wz3?!RinTAK4a=o&v zDK+4FwQ}6)b&JKmt|cXSjS8slGb9rp6sztopZXmvOWn0}PSCvv*d&Da$o`#K{XpPVvkdN{Cr z=$K7-j*lpeeV&p)?!3|a5@*e9c78>7yMs?V@0?+PYl-57Yo(hnudnp`hHmV2Q`EZ? z1v?5(8lC639nwdRX`AJ1v^#D6IH!5uYT;W_|FpdDqpm%v_iDni^RKf`QY~B$s{By) zd97#F+37`}$0EPe6fjBj1gKHb#_#3H!^ZdztkL8X6EoLga9oc&V^pM~;=l0Q0)v^YeV zMJ3F^B{s(^nb+3?O;A2(h=*nVG7QPLFqC_etohJoL0_A}VO5W+%aK=2(-P-D%Y*%^ z=So!4^sr|4C6*DFdi8{fv=0mG#L_<>+awTn7qZmah{=Ql)<#{Ad&-t>^ZItzCf53( zUJ8tuu9mO#ex9sz?wVszZnJMV#jph79ecPIb_$I(B6n`@d4zuxDP3)z75PXSlldH- zEu?akrI!og9J32F&}_(SZVfIO4R`v0f>q5j*|ZA>SD*NPtXzY0F*Qz?O`ax5yl$QZ zW+MGb%G0{mGZ}roC;=-qXf=P9+w=nuWgqRq)Hyf`(jLH=6g{7De3t5rU{LVp$=Tdv zWwBBxuD;3fbHSL8gk%J1wUUPcn2GMo2U24VIHlmxvs?bjof9SgdusAW2Iou z9knMJ;1&Tlgl3laOU*uNF8wfhoa!` zn5Q!bNtJ0q^hC?kv|8%(0~JoNr(8?={f$r_U{&Wg49t^1oFymnB#>5SJG zH%oM8I6rLoJi#|Yz~RaumXnnJsC>m}-1s4@W2$7NPOO#BdO#L4S<+Ja&Q{dyj>l7~ zGjUOo13Wr&%tB6MV%6b)1c9zKZm7x9zm1Fxh_&f~C`W78yZUYZ*G(TG4prK|=TbD( zz@X0OEW0v*v(}wTETZ)j;(qH+82*)Bi{Hi(ZzUWsLIpX!?G{{#vz{1=&X(6jyUG=- z;3?P%)6*IG%L6Z2kY?P32;R_ZJbHo;!#k2GMLLG*j%n*gWk0<=v+goi>uQO2wQcDh z*NVNhNy8UEonZg1POZE_cH9fud0zb6%A?B`#7_x%+S=FlE2hoK$m@5+Ls-F1MG?;P zw)b_9h37xvWuhjsm*`avr_wc|uV2>NjwWXXA5rYx_eyocB61kZtZhoP0H`hTmUK_$ z2(#{tSDPB&S;e~B8?Kj^;obGIU;e~vC0xxgWYR~~eIf|Z)}Z7az6)KFa5(qm)-i98 zoOYUUz%M}y*jCpdU~^Y};P)jM@Hob9!5EB&dGB|`s7nrLhFS@7#tmj6d5%v zDW-U(^TmG>vI94CCOSO98IrQIsr!>xtVxA-bAApLLs$E1;XQiAIw@|lp@+?B{tUlb z+P}CGgVad3V9%dsXB7V3^Q7lYE(^w;jo~^sI$En>K9Xf{2hvBrx*r{5r*|!-zHMqK z&R9DF=^LdU(ZPphj>sI9)w#$3>Ka|)>#Xy|)PHT*b-<%8_wL$zy~wBxa{DL=H|68d zdPt2UKt9Dci}aO`?-V%gT>J^%wmFLMHj{31nwq!nV;vExFd*+h04Vu6wGQ%rRpHMl z4gu7B6wL#rCBgpdfu%kyRSFPIF%REN8u)ypmy7X}_DezGNYKD+%LsB~e8pRPu$@9C zW9fZO`2$Q$_~?viE!kV-+CDNo4%%CEHZIW;Qc%OYcC4QMPU_GbbD(XRcf~1Yl_LeX zgVlPbzVF_v8N2cahZXGXqXUtunW!~l13@4Nz(%<3=A__Sy**NJCi#AtbomUCsO=|B zeXI?z>f@l z)qK^KK)&s-HkC|ZdB|x0nP2(7lDWJloMEb65Nt$v=FdyQ5ZDNmc7Ys^fCuXgrXCDn zvU;h*X)L1j1PF-7N5nt^z--*VtSI9X-9DzOu^ZmpOBFEJ{YH*Z_F+z?Inel1>F`fK zf{@6I0ZA`nr zntTZ8czhY$cUS6yg-T*sx{$C&P*jA;Dw8C`dUE39rdAns4qfl+RXFv1?TV1)xeuDO zwOCtF!0Q(=iafWHQ`?3*Zec2&=aSkDiiy3BdUwW=n6Yi;7S6aww_nI&oX%H)$@(H44u~KM+^#5(yDb}FV#KQ) z7wVb4DhI(>$9~Z0o$=KyNst9Q*B9&XC`HJA^XJl% z-}SP`NV{Uf?RC>6$>>ppXE$6_EI_SRi{X6!X1 zfYnvhJ`=}GC%`=#onHNy0B|qM^s8vJ55@=`bt3PE3qxwZid>0 zzog=*VYH<+>!qsp;$;&fS9*`>6Mf_HelqByGK0@e7ji$@Tr-})pkV!8DVQH2Ri)oR zhX61_3vFm%QPS1rt1fuWsbdbh>A)KY!&*nVcKbMfepR}%IpiM<6%n3&vdtSNhAHuR zdMV=S(P!cUeEkgFh~QKz;ExW4*@3sP&f$t5bR|_Np{SPQdQB)BCxP>Sr{%RG{=)0Y zzi0SoB)yLc0mGN(Gje<kz;`upQ{=+zk&?p z(Udt@^boQ&Y}cA+4ruS=uZVf1(W7J~0xWsxe?auCNrog~D)Jrmc5_#LjLeG5Ej&+; zl5W&QXeRd5&)+NR{{IjaiCOxWBK)6$62Wm@n4bkczDp>Q1rj6HLX)xm-3tAz_(61l52P`4z|xeY1*h z(r-F$-d}xdgy=E`me8!-<#^}~V(T+B%6CLg`?kiZ;axG{aVr>g+hL;ySh8#Fu1y!P zepZJ_DAJ=(Cy3t(CAs}uNVLlFpkVdi0<3g_^n}CCwgf?i^eeumZ` zWOi`V=E+}ZMAFIHcfQ=u%lu+g@1`j5m;Qxv?&DeYx6cxyc%Gpog6O;;rH>Jakoy<4&ESl&(2P18@sFqc z9|hASJ8x}wCA80e9YB?XW=={YDNB=kN}%=V6WzRBnWww-I=$YRhH*JXq99LbI#`sI zv|o%mNR2!|f)f$^(G@+e(WTCUXWpHx+T5= z3|o>ekIn!U;-8+ic~(R(IFM;4j9I?rMg1N^W-Y zEUP=H^MXupA-V^(@p37^wA~u-dqu$A`VcrLAJ*eLVlC~_;H;hpw3abE|DOP|LF-e{&&{ZP1j8m+|R7MCIv3 z$LTtP5E}TOa+;8Crf+U$T3T-E9B-x*Z>H;R2u1(hM96P+e+NhsY$f-4VCe$@ux9>u z5*hd*J^=u{00W)-&jYh|S}8M^b<&D9F!$2>M55kteYf#P`~1kpIg-*w;hG+ovpQ4U zB4<0fxfy5>Xy%kr@gMnZBNVoKs2SBOm}1~jk!vrMa>WY(jCwctkpKWoa5UgK6$t$G zC9foai5O1wot2DA^S34xAOZ#wKbE3neH;6Ka+T_*@LUKF4a6p-_)&DnCF&v$c6ovK zQfxW%#;Bn-P96ugctZ9^Nt$040yVg4cCVJERCe0r)KAjX{k&5zO@Fb%C5p*_2tSG$ zT^2~;??%Npr)AZp)lD5C_9l8{SH5r2&@!Ngn3&PqfdbkrhuMHX3y9H}gC#W(n9$_> z&yaCl>(e-UBmZW_!5g0kl)QU5%R;oxDGtmg+q@)4aWx`LN=6kt#=G|@*FSlt^@?SZ zvl&oIoOkJnTCZB3O$Z;R498z|lq;dmmT2>COQ^zb@~4f;_zT=2W4)+eKV|DOOm#G3 z(Y`zpe_auN32T-P=goBZH#PP5B5jXm@ZRm&g{_^(Wi`lJCuv0tqEflLm0L;r%miNlar7+vRwmcXY%Y5BtNf!7w5%)XIHBp^%`Cw zRk|oR^(KGyDhVrIo{U{*k#-JT>)Y@CQQNx0;aXWsXwyCyqYdRX^YAA^W$V^-`l;x{~(W^h2YIv%Q|Ym@;LBru^bB?!wi^$ZA#e$I`0)F-q#Lj zxmPkamgc!$pVXQUVV$G|&8d~pO(XH{u}hswBbemnM5$e<#dsCfS7VB@~L3+lnt zgoE=44UOS@^ocINO0yqD9!{SorlhvEaEiP-A12u+1)ufyf%o(F_jMcTe|AeQwEO@>Fx?z9D(r>g@^-xtUYYR)Y-7rUi{q4(dT`{l7n0RYnSXZ7WMY zBXfOXQlCKU0lb%>+ z(M{1=f6e;oh4tAd!uQD7I2l^^a#!yFbb{0Kj;ft#PRi(gz*b8`wq4Q9WfV9Lb(*oo z;-vq0sqC^;DOu5Wdd=V3#7_j=e8(+bH=^JtRS|5!mvzDTQ@dJ7pNT5G*`r9NvWGj} zpCaak-osI40VFK&g3_vxiaMhnDvN|;DC9TUX3MQ1<@Nsl4+(C zp_yG;^Luf`Z<9)&VL8-TF35>kpLZ&g;I#IRNuAFijsWQI(&yOBEBU{fmze@Mzz3&7 zAThxE?f+SGgQX@mS3P9^AP%xB!2-Za=b{JaQGw_E6^rM`d236P6kx!+3}bBZrOEUu zEKHFThu;H&iM^;-8P0Y}Iu?Hap4OO|5Az8+znWdZO}pZF-r7i~YbK;)MSpo#)YQ{P z(|Y^E-eL%&BLvHV&@QSUa|g6En!|JhXyHT|Xd`?}*ztfy6fCvwbxgf>dXNYJE&kaW z`{*f9r8yc6FoM?In&R>cSLKZaLtR!mel=Ytx8XO`x0D$`U?_#p`AGk~q>@j-le@wI zZ74uUDpO-{eiDdEA100>+-V%`quwHC+`1Gp-6_( z_4#!!DuZS4=DZ#vqr~NG->nmsiZkmGNgYs#ulB_SH8?QF9u2am;>djeX6z=UQ6-sb|$D#7D9Ndq} zbg|$OAjg~N^d88YWU@O{jydHw3(U8yxbtePIQTK9|H#vAoOveK6SiYM978 zZ!cdH>lPX|JIU!A*~-Q_dT*F+7UhF%`SlB=M>s__p&z{9%Yqxh>6D#Hht8I%%{j+l_*ZZq?JC~&v%i1}QswZFWg!&WjIL_groPJvH#xC(Pi0|rP1>=O zaI{P5!kzQm*xk1w<{d~Yw>Yc2@(kcXDv(LX8As)Cy*WFlUz>@Z&U02(2|7gsG(?@p ztgTMS!!=UDX)oggtL0IPCRQLGHH*T~-@RADo`6$$daCSJ!l7!W%c?3&r$|DMCWmF- ze?I&IO06DqFA_U^dn(2DiYVkU0AzwTj$gjo`8Ikn1@1k!;*bdp<@GBn!z|p@#3~wH zxnCOvK2E!smMybKJ_hI3LHE9umwHOx;-&<}FP(?D1bYBH-lJZ9=0K1}{+8F;1HmMU zBRr@P!bI?b38IF}Vn5G&uN7gWK(MDKb{|huH#OaX-Pgpjh687tXeSacyzv*{-gvJa z@T2bB)Djb-$AMyWiV+HL{apO(P!Az<^O{uZqw2!JH7Fr#|G2jp0!pAy=7JV_c0=5e zcio7=j}8?clTmS#f@thXp?Sn!aMZQD8JSRunuvoHo>LqQP9OqC+g&jUz{f6Por}|i zh8=f;sdN57DmbVR`b2QV`pK)yk{o{5OOD|u>@#HS5hcd?PJeQV1uzJcXFv(eS&D$LW-7k_~iZoM;O4#{^p`d!0tPS*+dtGVtKGggG~r zI5F^0lBj_Pe!?}t_Q-7KlbhvI>Nf-8p*)7qrgmQHbQxXb=o8LT$9L~-K@~$-Hd3Tf zjAS4KU52jQZ8@Bd1kvfm6WWpAQg;I*!GqbFq&>f~N*Q=1qqzf_r{?Wwm|L$hUGp5= z%>_bk7A`E?A?iOnu%QHfEJ|y>LlOkFgJ`_|@(oebrTW5flVT^T@*C!rqT6iWJP{2G zCA=DrrV;@>gB7C307sG$qdJ@d*$#5q1ujQEE-9H6G3P4i1#kL@|Eh%@ZWzm)meZu% z34=W{W_&)xUY0hZ#LgBmno{U06_^LuzEwKFN;N6z||@Zl~n?lrj-3m#*S`=XKt8mi^ziW0^(nK6+1J=*oea?v0 z!}%*{LTOhjy!4T*uxzI-*=TcZym)~i|6#~=*tBaI zXv%81_1&v))O9Lq@uS}%fr$wdmi3r8p&_kksHY)o78uz4-M>JBkDPCjyrv!Z>&@Ac z#iD?j=bkLgM=wR21!3pKK(?&~MBX7sb6rPO3OOa%{~>i_LyS*8@4C!ndAlOB*Z6VJ zRb8S40908ga{+q21oh@WVIYlFShAQEJ%8Wp$I;>iz|m&TM}HY=8aS+?;YEIYI6Xrz zHUKfh>d?s>_|`|@KoyhNNXGSrul#(0Wor&xn$S!%nOP%h|BYB4iTcm^*Fy@*WgjB- zk;{Cf+Iyc%L7}|JNdI0QOqKbE^OBoIb}o(xzo0GG@4Z4pt77}jaLipog&Z{~P*PNq z=dFtjrQU)%sKn2lCqnovlksi;KG>@&&4`IT&j~ z+xJ?i?s?U}-y*ssQj~kJc}I*iUzZnDQB-KMr|<~s88A%u!9$W5BoB)Intgmm<5lC_ zEIRYW)~g`5kI&3$?-wA8E<>sFtngcHR+=9XoI1}HjSF@Im@s*8h;7q_e$ur{ILio> zP5Hd&sSK#re^2*<3N)7G0Nw#Z1JooC12z596Fl$D99U8tc~h>8Df)OI4J;qtlmh51 zw4I*(ohF+-*1K>m+pwi%HX*aUkNmvp`4k$S%i1ME40P=lh}Qw`sJ-V8VEF}Qh%lgO z&7o73gL5;!9pO7$Eq3tJghq}KGcVAW4&#&KkxUWM-r0+u5_@MKeS*>FA+5LkaW`@9 zjpn>ao%zLL67>~`m6xl{D@qBm2uc zp)5RG5?s>m>9JQu?kn=vKg8*XsX-<~@HX=J<&W74ZEUqLvghlM)^@bfZv*L3VJiFwi2BiDmqn_7N_|Tj4wOp7#>@<3iiHF-?02xC4WAJNC z>-B3RfEjC~VCgTj!FoQv43pwda`{^}sV;bv{;w-6WY%CUWWxSh|21Z1X(a%48%{8; zIAx{6>LdHA!fo>f?p~;bl6>;tjZak%6{G`$0o)p(?C+Qz{nIhh{k3Km33XM@xv;#z z^WPP{yWvd0X|@_xp*I}<8B1#U?2>f@Y1W0Z za^@DG@^7>!>WSP8Yn{d*3_o)4WPIvc8RU@#=sVAR@O9&LR>v>Mq^bkWEDel%$4<nS6^|)+nu`7O*BS*69+4~3aX&Te#T?WUj3z@ zRjQ%Fw9{h76#`!Za`+MP<{i?C=YEyMs<80Oi<>k3aW?g<{pg67;#(ohzQyBw$Uk4B z0e$kkDuCji8YX{fLwn@wN^jn&-pSD!v75S1hYZ~AH$LYt#qZ3A9SVu@CvUEUIDc)u zYq^w0Wvq>-Qi?f)`@sN1tZ&{gtjtJxVY@7d&jp$%`an z=yo>W#lGQ=4^hzUnio62N?Dw4T^uhEm(6x8TqpBwbV)C);5;Ree)JZ8Z?8x7R7*E? zf-9+ly`!DaO_1kf4NtYRG4r3$jn?|FBj}7izY#y`c>g24gf7F#h~`3NGvJZ@78@H}IV@e!o0ER`B2*FP65r02BlK_ulVtisiu*gh$WAgjz`#FGFo) zJ;bzc@t$7|%9s&_)H9=O+u(-UlIbz*6b=bdhl?2IHB;l-?#*8JlL#d&D(WunekV08 zoJ!~H-HfVUN2Oi^+hj6z$At-W}GiT%7k~nEry(Y4l%!>g&^b z77J=Sa<;c5hhh&Mh81DNDFuT6pnBL6wk)={9c@LNCJsa4VNXR%Z1-}g{jd30d@HcK zJ0xR+4@T;jdPs+k&YD5H|jGV^s$oTjhy zRx9!_lWm!9OmZWiT43m5UJ{;SlehV&d`9Cfv2{-53Y7^6`}}mQzS~B{{{X0;YLd_4 z?)42f?RW}QcnGE61Ux!butPS%%k5URr#U3Xm*UPdKh-0Cq}Li)k}8@Czk`ybXu)PR zsJ3=)EWKqq)@Lj3Lzg^z@n2#Mz8=Q;eR-}LCl9X9h6dBS{^(>+6YQFMtL@u-0O0JW zXrXH{N3`BA-CYhN^MriNV6$wyoTgBn`Qdcs0Hbly0XVWh@55HhXY7Ei!4fID(J%f~ zn}ZPF=jwRR>s~WW@+7a*jORvoEVSN~gzz36ZXLL=@%zJE;hA~^76g}*2KfAbM}-?p ze}e`k=wi^iQ4nJL=@=gad^yk7~-rma$h+ChhLJ^6KG^CzZvXkyyz8gB;Q{pUSW&8SNB zzxlCiwQ8c2`)kbX&PGY$dyS}t@psVhC47P-5rQA@m_}EUD681w5IFJ)18jw!I!1}R zrvW{0(%JsW-u2^hssnHaiaPIEMnmbzFliW{Q-qpv&o)yGz{30ZuBi-cBs&_4^Z1rc zF7}EvgBFcIPxdR}%=(~qRI6G@Kkb7SqOLff!i&`zvObpdDal~J;EBLBK1Bw2z}}MM z!|!wRP=NZ1@X(J5R_Q=G3%0(w8!xLqXjk&$UM4+Wi$q8noqXnTN_hUXSC7L(JkN2S zBFNxZ#`8q+$8Y0+UIXHdJ74r5agO78Km{3-9`9Rs>lB0mqyx|+isY8zDr0q_;BkYR z8>3PQQg}0EdoP5n?Fag4QQeK;+Yy%{kg{iSUs^h0$mA^^5Fmw<%*0=Nog*d#zhz0; zXKt^+(fWa1dRQ*@z89q7r`4CuH~2~2vK8VzV-j4Xa}&pTJKGV`1XKITmvNC=oz~}} zRgeZrrqjrq(sQbl>wEY33P{DyJn#E!Pg^iWSVos|2Iulk9ImN^sVwMEw`D(Wx*K&e zeA!EpHA^(W6f0*zF4l|*eKaW*n3kdaiWSRup=EahNt2;Z$9_a%I2s(cp7s!<(FhI zlT}*XQvJjMM2^S|qN=;*GS#U9e}lU6oJ1jhrWIIW-4Fp1dmiCD4U7v}$n-~3fe+OV zsMC~u`Bg%R;MY9wGaUDPVFbuVeOm8z?ErgfvO4s!T-+Wx;UkejRw~YG<9Dh+ANc-l zTHzd-S<MofxF=}Rae*Mu1jvkbwbB%x6o2oxz1-t-ju+w;Rt$t0`BkgExX^$K2 z(qKnJu5f0ngGf-Gra3=&oA!ahh!Hdbgk5ZVHpo#>@weZX&3PiJ9$S9?}TnzHH1AqW4% zkM1(@93s)UFRllF$I#Ca*5{H&D;4IgRMEXa$AfSIpsivulY0?C5km(*IxNN|B_0d#%(W_JLTa@+7{E92q zr~(Hjj}fr7d9uF)GK?#bG(G7XYOz#^Jd3~jtHBnfl4>Dl!j=`@{O6M{U*L1O_ozQs zQJmu2L?yZ(B?*2c_OU|J_&5@luu!hOhbt#G3>&no7=e}Raqkrf7VEy_;Xn(e{as*Q3~bBV^I%{ z6c%z@(w6k}GO#Jd%(natrb!@2YiD0R<7QRe8msI;DzkJ#QS2)Hd8d9{4WsxXX=e>U z3d2#!pO3$vwtvtLP#{_RsFR@3AeU}VDYBK2rN0oso30nt7RePk+L%ZV3*~ez2JS8> zBkKldnpysQC7C{RzLkS_@Vxs*i0>@0uO&@(3_3|HUQu?L%7{qV>#~#DAD`AneYjKl zNOuC89fjqxA@ztsIkBz3YspwBo1x6Q#FzNyq@BaKlqgzIf&mJzm0_0>>i@>Js#Ila~+1LtU` z>6J)mK=@0uLP99CWg1RxaQ6p@hBKcrvb8yF(qm7(tMbH$E29RE@4)FJ!SGhUM@axN zO3Io!y%cK+hQHgK$|^A@R3-}Zb|BIE*cTr!i-XcgvG=>fYt*(YYE$ee$Fa}J#MQrB zRbEfdIS6b$EZ^5z{gIBktvyuf&Gv{r?SrxvxP)`cy)LP2JFbxpoV(7Y&M$fW?N)0ZmxXHhB8aYdyO7 zFDa2K7rz#Gn*IeSwi=hlY5GgEx>0U(g!)s7D}RZd+#HPSG*T{!{;n^*{sG_ynelSC z_r{{+D+%)^(e7u1nt)>WQu5rQOVNh7hvG!AL2c4PHr#ag)SuzDpGDL~C4U8i_4xu* zG$dYsUdsFzHsmgVFgbaf%2USK`|-h-HKd$~0AfT+7^F{4Q?syMuSK->sRJW^fcU?; z09okS`ZeSeJ;&ScF=mCPS9%PszjIlR-?6E?yiR`jbrO)ZBO#?&qLnoI9-~Df97ZG< z!Ye9D)`hxtpp-kfuQR0rP0Sx~6<9Mn@@JMnr9Vc~qn(6=zkOC6#UIdt7@`$Qd1>#?G9Tvm0|B(n!L&am} zAQFN08WKH0(q|!6X6xLc^yKKzc<-1OKVRo1hrubekvC9L@g((_qlu)!$pXmFEi_x- zMafUq7oD&Bp2@pSEV*4)26>lWDKejj(7J7{=_=_qJ(~1f`$)G0+_OA=wo*>P@HeCV zsAn@#yB;U26)qY;>S0qB70NJX&pmRsImJ-&^Aaorq};4f?*ri8(AWeFLhGwSdh_a! z4)aTSgy3me{lQu27;eeqbX?x|D7XC%ztOzOMkQZIvXeuzuG*A6>#&TB7K7;7FGoZZ z%Y(`>G#qZ@hgA!_$cyf8c#n_22fWwsUANpRj@tDMxYk$MCVzpGxGen@C>RtOgAs~W zk9kE_G$-l6IQ|L6mP-A8v(qaAajPr7mSl;z)~N3@3%4aizQ+vf ze2BEuBN>k0K%_}-qe3qg&E1(=n9Q#-pY?nXFW&Yw(5PzAO4Esak+`<|wn&C`tQ$Wg zkaCUh&cTQP_)tMELxZu<^R4?;i!*QjIlhOgX-6efe;#W=YjMF57$A=@Ad@$5q)2v7 z%0tf946b3V`MRw_Z||JBl0cgCf7m3szf$-6JN^wwKt*f*lfMSzz{;r~P(LX{H>a{% zso*38K`SZIN#A;vbY0j9@)N}5`WS_{1=wy&MH{b&WW zp+EQb;5jD3_Fk$n11ecxCV$O7cmivUR@|Vnk|(P6^CGjVyAHe;OTBul=&rMUH1^5^ zxrabDg=o%otKPT8gryJf%r1zr2hVnNlVjSmrp!1(pVNZhzr)%r4_VVSOw5}NEDl-I zlDxZWZOI%(ugP1;zACbKqF|9;MxoKj$fLC?&kVnoT8kiWZ<;D&kAp^`4XHdF%-erq z`FSGvk7qcKDLG%e5R4=(m2qw+3kQ|JBo2E+voLdtU$W}&T@+U^AP~Qi;1n7AtB=ns z$QMhybEvjT(5ez$6I-*1Ur*V~888Fr7|iy&o(H-LddEQhEB98%GKTeMx%X8feJPrC z|0q;hYgJ;d9-JAFmv-eX0lTO?*jGGD-y>Vwe#2AlQ+ZtQxDKCim$7}Ci|>fm9R_C- z%hUFGZ{9EYWFB1mPuxp#Bo2gKU9P`B`jErX(tz~MH_UTdCGfvn%R_vK$Snzo)}1M^ z08KV^SPr|i?$eCBoZEQqTpk3@poIS(;T}^CRwe1EgIF=FBHHlU>HCc6Ic-?l?ZD$FhQjg`L)fvC%)i8Nn zrmyo%JA_zSuGh6funby->G*h{-I2o?VY-iLCVO_$wDsGZ%Oq`Ox6n!oW~hbaRq`dJ zZa2C2Ak3SxB4v|gb$;F2P{-EH@%W6Z#of?1qRs-RroyiJvllMwoj zyZeG3a;r6ezNpAlt5R$&WDzS@|NT*TQT=sq^Q(r0)p$Z?q`H5Q!#r$P>;w8lMPz0; zKQp-(#hpT>&Vk70Z2AY~f*l8ZQR;bv|B~XQQo3V0pXhXOODsvg;VXF7_C=%;q7@}? zPzH?t0^!$p&EQEHw-gXYhNQZDLGe?Y=@*iyPc9Ap@#cEu1|#VoX}qOKx|L-?QxYpT z1wFV`8b~Kz66<&GN3_($Kc6L~n*Fs?SnyTV6|R`9Pr@wVV^h{jcbbIPc-_@vl%C?x z)BqzX3$w>U;1gEYDvmz&qNkpMejGm}4p-KWqYaqza`Ov3Z&N&WL;B1SxR5^Ckk-xA$+|179*8MH>0)gJh{BCU~FNU1-{O4VSP_jm1BD$ zy!Kh+q9rR#_b>KJxn^JmI-Kr7>mX-FHXzZg{PULL5I*X6D6lL9ZfnAHsNiz}98^S@ z(wUiZ;A}n1=(5svs`0V=lRnX5xV{D^pETb`b2+F!ncLC+CJ_p(ttp&e8uK86LmD?L zsPbystb=s-`OPQ(i|eFN)O8|lYu>FQ6>q{?wE#|`PeUyz2v)V(uAX_Ul$g`%W7X2l% z8pPb(AH?I)cuVIpjI;7sB|n^pCIE7^GCIowyT+$<%!QnUH^AS%E{7yARiS~%u*f6L6sITL~eAsR6fV?9)aa>1^KQR6}Y6tKmg>a7`+NgL%ptRg9sa)b5NtNV`UNQQshT7xkt`% z8>Y@-+6iHrNz<+z_sA_1rhQEOymy$aw1QZYKbxo7?@{puTD<0Fcc<1_OSfV2;0ZkQ zYscMeF%p)~0s(d6z>;PX=A(48lD?Ow^u18jV_KUU`iwTF30^Bdch7_6?)*NRcfO9Y z@Lkdbx(tK-KcfQDbgi>=+h>OpPJ|E(oY{iYWHr%)A#UxxxWQ2qU_(rr`<=G zmv;>4Fcb|}>$^TQ<#KWzoXKgItJL=F{F8_8@l32Xo3Kie34BGubc{GB?9%@X?ug67>yRfCHm;(VNiz`ZO}*qsFXM!f3r^?WovV`cc|mjR+b zS1yI2@X3Vn*A@Gq;0%vK4Luo%LHTpZE7gcZw*|ctQSF?=CBlNn<6_xhy~X{x@cQEw zrUK$UGjOA3h8A?RVIS_ASie+POy4`uS3HbaNF%czX>xXVbnbNOlTy zG$#JTl;Qc-moM9AYv|Io*n2LtL-+WY2P8x1GeKswruU2L#IPq8b#mF#`Z5$ylu&2D z`hqjU?_*hECzSoN{)_gYpTb|k)9_u6!-L{SG}oFI-lsvp>arOt)}YO&xtgx<=v0U4 z#x&Gz+4Wmy6FZ=-QRk8XtteD`4h0yTC99w zAaxtpv-yfV^e7l@!d4@ah4~R|B`0rsn%-n^q^B(z^nG?d+%MNH{pPVs&`J!igix`6 zF29##m=Inl>Z=Plj02XE?vZUT>&L!Wp~!t6IEyr_Md4ozKb~zrS-qGBF*-kuwMSa( zy_E3JZbdDBYHo9rB5T<QZ#820-N_KYMTI<`#Ol#5VLfu;?XBsND*zw z!yfhb7X;#1YQp{oVK89500F79ts( z+ITI~{{ps6RK`~o&BnOY1arFMEgs|^~Z2^Sf1-cR6 zNsN4*S&p({Ge1&zLmS~!`T?yUEZb>Q_k2nz-n6SHOIwtM1bSSOxE*86`a%Gituqp8 zW$Z|QxH|MhqTd?1`DA@4=;ifCTVie?Qf_aud#io?$~N9%*n_SvI&XBzfw(-_aATeM z^n=OE3TqO_S<7H_)19E)&b=x5z*2O4@3&8lT3)<|jg4JW1g@Fo5BX1r?4rAVD+*$oJ8_Tbn{L);I?jnXg^B%f8AVyg-}T-3d<51;|H9sz2&8GV zG~aeZ?>j7COwoTP7r)yvwrpuwn3;W$++thWnGh_oPk_nB5jQCY!J_yZ-cavI1=me8~yMZ zBI6Whh@hT6NF2rCl0kXxUDr6YopgyWB1S&OCqMPb_Y{d^*5MP5W=^6jEtIY3_WuX1 znC`{R+a|A7usc%5LoG@E@)~JmEYca_U!|;v1JU2;vgqx*bk{5;|KGcW- zr;54OpwB)zz5t>0#C+??`x8=%jViiZtVZ|)5VwO@(mV6?dc@h`Y#a*@fxh1MxaI2Q z{xm#}#Q#y;fr2#X9=l&auF%nAKb0aY)6UQ zSye#f3xr(5_k}~g!0tR#DKq6~N0pG_j$xe(wE1M%xlHBymM;M3ruGnatRK+3C%R?I z3g%Gzz%&j4Lu-Sk)26p@ssRl85IHp^>7^?xd&{}wbJZ$sBrD(E#SYUPzaRdpc%`7c z_Hxnb{keqV%n3yVa>{R2fJOq~wDXI-_bEsEXZvi*UZ~xvrV-hI zNIe5gbd#8{6TPe!Fx-A7x~)DWaz!8vT&W{2_(v(tF2K{aKQJWYi9N(ft9t83DFv4J zT@p%bXa!5?K)Llj`~~V`%~c`4>!~i#y26LSdBsy(y*2eL(|&JKb;bt#<2gh2LH(VN zY2dX|M{z#tV{yeUC$7jJnDgmFzW-Z3+!9C!BxA&= z4CYYzc*{lb&CIRfxb0ihIT{RU`lQNnp-QejeKo;jLp~w|Zdeop9 z`@xeqRWpUY^x)h?nXLD=(DKkl;kHr6!`0IW*MoWncme*P@;fbhQ;J#Z+*n$~pi=U1v*zR{4KJ{o z4DQiTJCMn9q&Uo}IS;mt(^va9P@~p%+{3!k&(#y!e3zE2QS}*Nq)8Wde2qaedx`nr?BfzSn?@LPY~@#yfLSd~zDV00`u!+IJBYcN0n%8kZLeAiG(N{OiHuj1GU& zEb9X6@FU>fC(51GpL=I^Vs?}STr%-0{#ZVV7hl21V77OdZ#1o>E4`j4^@RrMM&lZf zQoA3S&1rSxR@XJ5mTPi9CYwgFy7J-IH41dq+r{3nP%@7ogzga5R3Qm4XH?oOfj`*9 z#i7D)3|0qIjAz0I>>bJkYkovt?+_zBC`A~vD(0#gr3vsObN*rFc0cM_;=(UOx5C|) z_UkJDPC#kw9Wga<%TJ&wj zG(YuAJ_}oeWMN@HL6Nu=1Pbirm3zN_nPKk*?IQ{QKpnD^9*Yplg$0~mFlWvmeBhz} zreyKI{qaUqlX>o$L?f}Km)#;{?4>szr39YF%VAv8}^%>G}Q!5*(sblVdVTtv$#PPG0)! zxkE6s3&8QUN@S5FS%?#cb(_qM%KxO{Ii+R8!W&>OsTO2uuKk}+Ubdi_5pOP=i~X$D z;Uw*Rqu-gRMPw_dmBx%o-zRk$0kyQ$4Ps=r=AdsyIz<`@*oX3p(`xS1GIQ&5%FnXs z48*BUyyvm|wF1jTvuF2lytUKnenvXRJ-yI1!dq5f#`S$Hj3P|78(TsxE#L)vlwaiQ08xvRnG*0)?1_dF{&;la<{p=WNs%wLRN#IyG87z9R>9NFI9ozG49ZSj5{>+V)ht z)~Jr>7PsA8dKJc46aq6@s$w#WUQLGh0{2={Z8>}G>KSJs6!LiSS6(__^d3j2ADA0k z{-JJnBXguasUR`muFzn)AOP)_^DbK203MqK@@f*G4|&aN-CuKKXDZ*VEde+e>uEc@ z&G9La>AkW{Xv|rG)#1f-@j3}bSF$rbHKw;FQS{p&7t_o~S+wBLb(KtQz4gOu_i|tM zwPJU!xa`OXTgsfwcap{(N~%5p=1y$O{<;+3SxbiV`0K!0GPlc;;@u}9D+#xzh#Jgb zDpO<^uCJ5mvi=oTat%y>&*+p_e9{8Dy-hc>{>9qBcNT92yur_2FYMu%!s%nS)t^DT zAm_t$Asj)p9m8j$BB%BI(A__6%&Mohj8#Voo!kN=ZLtF^H zG+a-R+mz-I+&g0adq#GBV<5O@chj=YbF(9nYvXWaxvTyNS`4>1xxkBu`C+FTSduUP zaGv;5@7YHIAxZ@7CHqjh~o3PJVCKQ+}OA;LRSi} z$~||UcX@jI;Fuei<0pgAnAljK))g3p1f4iO+A2?&bxW1*DokDvg+0>RDCF{IKYSiV ztDr!32=?JqTynYpC%E#2f-$_p9(m%_kb+#=r(je(!ZG^f`7N=gDdLg^Opy54^JWJH zv`g*5Co$#-{$$Qtl<-PL=`%F;k9**m|JZyddE{Zr`r#Sg^J-|l%9XYBT1qCj;tfmk zT$nWQlm~iqsF&0FYY#7zJcyO_bnk%pPvN}z6!RYp&Ew6X-M4zr1=4sbx+&ZHVvq!) zB173FD)OGd8_m80S6wLoo*ZmJY$-{(J=?{P`CeVw0bE3hrMXYTqRpw3meajqfT*)rcEw zG|%*4Y7|q|ThKerFRd&^Grl1PH!oTI%<7}dda|4O4pQk2=L)Gsm4#{bkC#}!7C&|; zDq?h78>)_k;=UP#5nj5)bBr9OdoX8CJ}E`cm9Q-{J6YQ`RhPaIJ5WnhahH9v!m(i; zjSe2FE)Zlp@*0~Fye{H+_*EClsCuY*RH(UXouC8Zg4rirCkYnb8mLI4z`Q|>;s}fo z(HJ2mD-n>>DpisnvqzK;wpk-|?r^g8V(U?T*lp5UbC|D5jo)=@riUm%CL3!RUHO5y zj=5*9AU^Y=ven~QOSkH>*StYg3+DNKZPNv@(eQbSC*=0_xy8q-vn*M~!TCLrqA!N` zLF$1s7oRPzcVAdu{2g=(oS@$ye=*|vWryni=^02?Tnw@@{PIk3HnUG%3WU*$9KwV28?AI&< zlf7UMpwHN9<~yCqa=hfs+U|7=Ow|TVeEfRQWcjKCggrACtY!?RG8zfkHg#1F_!ddUd|^jBvmzKI=~xe^ z%X0O+b7jTy=x?EsT|Y47LWB@gygD}o&Ha_b!XMM9se@!*`Z@K-w3uDG0f`jtHQcmM ztTpT_UJS#&Bl@aaTI+uX1rB;8UGhU7gpv)6DjadWf~Z=WEyz)dncxa$!(i3}^JRiV z5ln$k$xhhMBxz06A2;VEhB+ocr7jma>w0LrMaLQ2Fzvgfr-ZKG?fEw7Tl@htxuAfp zU~6iMKy5Kks3~b5x)|Waa(7lk0oI0JNtR*^M7uuNOI>guGdfxjf>H`SWX83W$4@o7 zAnnOIEY+${nVH&tBC{fAyo&IYVSev6btEM0vEgO22i)eL#iBviyj|lppNACS@LE=> z>X{poKi23=b9HBPiM8mn@RJ`jiosRnZek@JWkaxux(7eM6g>`mKzPH0F@jPBH5a?w zp468+qHA0r1V0{C?Sd?vHDdsr^k)J(b`0*HUzRfs1Ge4x(hrCfZ(igs>eNw1LQNxO z6rSG>aM_ete<$`5iHF&pQJ+oknw!iH-f?ATA}^I7$+!walcT3OFq=^QhtWb|;*=O2 z=HNne$Duy3$(i?SQ~#g(-6dPRi(tkAXI%q5JE2!GowS&gYNx65Rsyp`(*_u)c{Ljf z?EjIUS(B|J+#}VbcqkiDY+OG!tOl*4(V6A+G-BYAn-GaTA^le$E*UeB$yF40%%Y-g zKE7Yi4R^WVJPt@9uX!I$nAoI2)A(c|?!9zHDLFVzNOP#L^64)s zbxXYR{b=K%4T?B0G|3KW%je=n8^XT4tR^jc?}*3vg0Bd!I$d?zgD(j@h zUe3waif2?5b+U#s|6o;>6+dC-KZ#nZK~+;`c}kwPg;C}0N=YZ){?;IUA*Ai40|{fC zmJIj<0Ni5w?|%V`yoB8)`hSv)oosHrxt*p#pt0*~A-BPdh|gvB!dfk~dRL`~5>Y-m z+nk6?n0A!KF}dUIPAEh4iXwsaCV!RV9@fmUHZ#?KR9Zmi$~>_yI@m;ovV$pbzSRbf zloULDRJ6em(z-nR-hZ&!*aehTgg4B|U6tu$Yf#hisN1X&*}Vm6rTVdR zX!kggE6|k0JOzu=G5z)kF0 z3s^e$A%(ru-g;dMW<{od6kL`6hW4mP)9l^Up&B5^)M4O$P4=(fchkpZKO0Lp6Ajym zoxJFz^^q5$XT|&f(n8|<(K&Ud#8x%;FD_{KE9R(cna2&k8mIpC6r#epvK)MW5+E-+ z@*=YT^S5H7d**nn+N$e&N?4cYyvUpE)#)Zm82@2vnf%1PSHM`FaE=wv?SD23fH+as znbcth#O0ltgqJ}y9CH1WM^jejd0|GhQqe@wx0)@vjaJIT z7*nden+Q{E^x~+BfMT~hXV^}XxWd_H_5&Ctt)H?h%Y?7U|JBWTg*CNp3m8MFQk5bg z5(R>ELFsIYAa0}yDBz}V6*#nn5&;8=5fJH3ItEb`6{G}^P6AlKAc#tl2m}nh6G#Xt zoT%sZe)nbHhkMrBnq!Q)<{V?LZ>}}JG5!He=e~^bpw-ZlVAT9^)qzLkWk$M8srcRW z94w~X#rCTa%CseD+(cC}uHi=XwV#-<~;4aepGFKrx@`fO z%#hv|+(o^_h<1<>S#LJuX5B;i9L1W>id4r#>-0BNQ!0l3avJh7e-fjDq=26d;e(ZTe#~sLywzFxTV!IgNgWLFYSU(Q zG8;R*SXumF>+1pjk83}UEUF4hy%TUj`biHQzwNLhQ~kX{-aXtb%_Z&L^q+zmyWfRl zwf3sECzUpMMr-ePEs(vK>C*l(mmUa?%M=Vl2>o{R;GLPzbN#H!P$dRo%CWfO8f5MJ zbIZY}$kq(yu6#O54wK{OPc5QymgK|1ly{mmNKT7MdS-w%NJDuoU)JZsA{i!-AsL>5 zT+o02qHDX&ZS~wa)$U;}2Jsx)|8&C5g46=_?$OYdW&6m;Y6s z)5HFgDTd$cejyUPUuT?uU7NwL4V^Q2nGjkcideBOPmDJF9_QXhPrxy;{+Wm^pP^;obs3EvsrXc66q#I2;GPqQ@9Gux z+!oaNd^d%2bA%~gU$QzfEt>fNl9mCCy78@1N2$C%r^J%l+?j%1LHBi+5RNX7grF-d zrs&;ZG03pq28Zf9nPBN;o#NFehSFN_o7eVhVSDX?s2CWWoq<@W1@u?UYjs!SJGi-4 zfkE5cdTpc0tHq-=@EmjTPo|c+Q{yM9>J|*l(tr=5LV zk}&jdXPz2~ecN^emipPrF!UVuQL#G;yYYIDhs!yGNZ|i(Z^W$qqh@^x2s9If& zE#i?T#^JO2DLm1QV^DRctwIpIQlGA%Q;X|BrbrK%s&#KuF`hD5ps8e)#5nU-e5b4w zxE^q}ffCm^257O6muZ|QUgu7_SJu?EjRxyrI)hU^8*o_#{~QrH7~8)gD(u}s!&%a3r3xrCI*bz$^=hK99Qe= ze25ikv9h}7L#4g=ZiLElnXk2I?M^Nihv}wupIU?jJ^bptlL|cqf+mvqk6+5QU`gaP zsyxUr5RMLc+Ow4Rb>I~uIkwTLxpMzS;^q-0<{mv>kyqY^;MQsi4*K=mSt}*S5r_bmwSTP$@nCREkXy#5U zCt&PPzsmyE&uX0g=OIL#u^l^eA!)|~c zHiZ}HnUa*Pg6Mh0#g}ofqi|Pqg2XGmv$Mr%1Yb|D6t$y@!`kzqkd3vSL|k6Zns~$C z$vJs6Lbn~+hcn_?c<|c_b<$I^l0!tNLoP$UI z$k1zcH6#9|+;E1W-E=dZm4bMq|FTAbot{&xJI)x$MiiaRs#6eYvFHDz(Mvqo^#zs- zyC+3R@WZ$t%IB_8lB6Z69||T;q_3Kcb@6obKj{V ztaC(!r#FqOf}ovg(za;r+nzE8 zHM{zgKXpqzv}QJ3P&inw{~5rh9}1V3`vlCG)DUS2*Cr)kWw-%j_W3y17kn*;L|Sg; zY>6L=#+wnBt$F#8dV;{3vghTSQ3NyMezF0(K?pd-;cGp!F^a;;fSxNeCXGd+4+AX| zu>>9f1mJRYubmULsk?b|awP69oeKZ~y1MSQ3xhVT>OwzGe~a6Pa|0mvn19Igm=VdkbjO~;g z0rVp;Y@LBp=QgE(LLMS$Gq;s*M9dku^cA*VVTmiM&p`!a&Lx0)A zSk&}OtWc6hDjW2nHe|~Z@idCZ-oXgtP&o$s+cOo8w)L}0&uKm(HJx)z2MFG z!p6o=)`ax;bfDQ=wV9y}Y{i)4Z%WA_byI(WHVsnt(giYg;PiDRL^mT zi|is#u|p&uOFE{!=C5fK}UE(<_F?8Wp}i zfG!Hp(_lBd@rL&&OtaZ63sK{%qocBjeG)Xfe1gS(O-@E3v}pl?Zaj{3CGA7J#&_Ra z#4#`QUij9o*;M0z?!W3IXE|5aCkEQ&WuI6dV7p9@d$E|K;!^K6^mWd%=j&#e66S|2tyGNz?fEpM5 zA41lvg2?>Sgb&Z0ZtXGer2KMxIGU7hR8UI~to&qp6?rzk?(=^l60Y%)v~F_}K_^|_ z80!>mnIiCgG!LHn+1Uy6vCI2#BDMo&XJv6@UaqRMeERqsZj;}EL>|mMt!LEtRDL=g zkl8$OaXBmg`N2=K|1Pp%32DJWIH|YYn{mmzi(%EZ&!kHh6ht8zC{L;qW}UrEU9qs* zkv5+Y_GYvleJpxGj)~tO;i+)zvh_K9uPdu6 z?lFKUSAO6W$YJu)SIzCWrupXC?o^Y@P&`VDwI(2brP!3kR!hbBlHl&n?d1K= ze0OGkbMBct=a1RF_v&U<)vDTy)v)&SR7Jd1mBU0MMFRi;m}3%LYa1+JY7Gq*4Y2k(mWe4&5D*6&=&veF{72=_(ExkJ zHNe`%xW$IJMOgcs|MqQz1)#w~upth1TKtj2H2z)<4T}wnu=Ba$`MKH0=X-kkVw~V& zoW)`T?1T;QSGWi}o)@2=m)hE%8!Vm|s0RX#!{!st`N`Zm_FsJ4Oe1bp$ zoJer+P?GfUJfHqh%;?LDp!7bsYJM69ZYgKC}1b)h*&TGVLM*YCo_KP^zjX?K2=CwML)`&m0;}EL=T1HeJ{k)IJ(avj;$HEfe5~ z8{O$Hatl$_)YoFazC=J`c!zVdSu2^)cj`a8uFC_x53PP4IxBcc^j7m*^*`CYETH`y zOZl8H2p+vUR6niXN^7ZSA9H=E6fLT*=k{nAB+VXF)p&ZEKJ+KrzA3!qm8=w6J$OjQ ziJTRi_sV&NBJAByd~@VN6%Q zYd>ugZlF=WLYv|$E9h#5GGXO%^ik6dFE7Mx;PR)Nm=XWmkKcsJJc$8tLCXjT92tvM zPwqOgWkg&Cp$G5!fJ042Z6#u_%sMch@+I2r13_mMdwD~^%bhXNWILyoJ1ZvdrUe`F zzC=w@4lie6=H?^6^P+R7ZSjGD889~G-$6l#kGiYgr_Q)>V7d#%(;?ca>%+H%P~#4N zH!G|t|A5&`hnM!yLVjJT>F5hmc+*hezi=V!) z(4;;&NDqdZW$L*5^ch!PSAZ>x%be?ODn>f)^=#)R++3?0lHFMOk|h!nH8Cx=J#1i8 zzQjaJS$sk8PdKT3?$^)voI4A6GB@Bd?`1N_yR*&OaxCl9q(hUc%!c-$X(o0^D|U2C z`{Ihi!jWwx{gCT+rV4EWd|$WxAcDov4jdJdb<8TtvhI57Ay>8()y2=o*bMB*I3l%q z%FI$VAaT>QZ`Cf^l2F7Ftp?1$9bP;+Gf@%J-vI2fYdM|@QGQ(cT0N&eq5y+i1U~qV zQ<-vfPA9moLK{hiGN9iQnc4DxWze5U4GfS{bIz$qtZn(W`}VpaC$w}XN#Cek-fuoX z6%tOizbn-@A%pC)b27}jKtj5w#7(L*L!qr3SXZ7ezsQu$v>+4t%}-Y5d7eyS?_Ryq zx`qZqIj*5-Q5VfA=xjo7nLiyMm{fk^I7yAIqSET;i4`@V>)HU`bF-JqmPOt&e(7^4 zkj(D%9|$!P;aea#&ur&Y^wzuT;r%ADkuueOJW+XA=pN zWoMHCg)cl^Ykx;{z^()tO44rLy z4Wh3$Jb#LQd5}JpnGFjc!zOE_pDX@O{_&i@GpbO#&`rsNMpQLk3Q@peQZ1|0u)#T zte*s}Il23Zn>>^?+DGrdVSOL%CV4bI%D6T^7O`L1c+~6=&!loBflKhQz4vvUI}VdZ zfJ7}0uk@B4HMtzqXMC40GAu(NieeHRTZrn zk_AumNA12?8>C0%$XHuX4WT^6Mh3(vzgvr+30cW~fi#>}1eL)TegH)KSRw%!houoI z!46+Ce=}3K1D4W@p#T6H4hf)N+yp=Z4|!pNfq)di^`CYO&%woJi2}-B>gBQpbpe|G zYgYg3asO%%J=yt3#R(%ILE$UJ+?gsU_8n4i8>+&ffE#h*rgN$4Uo@NmO6EsTZ#TCq zpXQsp@^y72$cr=*MSuGgy_~PO2{2WU+^g-qogMX*Wg#1Wzgq#xGuRDb?_5Kj{*hi< z-XEf8YF`gIbT)R}?~er^OPTjW+4Yvgu$XT3yzPRr0KR5+0Y`)QRdtP?P0UaZkZ8M)xiS&&q(DYfl)^+_HS*{eB4dKqRDyG?)uNVupx#4V3QOKWWc*$U{& zHNJ!$A&DW?=EgrL~S%4fRzYJ!w6^4zfKPM!?ztVjy*KVwxF& zr{VKj@s~BNu7ByFlFJ*DU<8_;SSEw7pW8{)D!1f9)w1KENK9hk;-)xcKNsci^+uYz zYC*}9+Mt47b7B;F$V!eD%?FV(J+b6TUQZ4YA zJ?5#{C;d%~GoB2Nr{etGL2D)G%Cpz)1d6P}ZZBV04v#^OMT|L~_C4+OgONP;;ND#( zFS5FlAi-|=p?RTG^@hHC#o{MX+Lx+PqvoEsx@atYs0CuiD?|oGOWfuAN1q>p5gB)s zYIHU@F^*`Tl_D1Ma?dR9ud2B84QXz7(d?#D?W=M*11|ywR_1l$D3iseeJ%Lp zqA|05?^QJln0%esFrcPkQqk$+{u|Gn&360R;&RS2hv?6@%$aab*0{D! zKWzJDmgv=JQ9^O?e)Md zT@0d-PCTj3_*_SiA|&J*zFYlolK12uezEc4cL__$Hf(j>+-Rc#cSQyY4Qm39=uEzt zM`(Cs8tc_ZMS61fYXT=P>N{7i+!ojUbc7OO&38&u~27N)-D!uEU4@ywzz)l18c~LD8*v zQLbp*Sn6W%VkmRI{s)t>-`g6AwPX{hv%5hFtw>=)(VX9@>z!5fS6O~+zih>lHJ@3G zV%d!NQkC2`h_7)R{2k*b3z*B*zyPfEA=s;9y$p;u?C!1Zs%-t?JWj_?Bg*e0*VKmf z3#O*$Aw6ZKJ!7vS%z<3;43C4afu-A?95zl#%IgKwvdZW_r!sKQgn{?j+rAH9-@rvqY`FW>56!$$D&ecvI%c@=FfHt4FOFAa|`2yO^==bj?FbElw_=Ed0$T4u< zy`#B3ltGOznweby%?7B-uAA&>*cfF+|B&FNNmv0T5!t=CaE5QP!a`h~vQ2+*=irivKSO%e1p z{{(sA6N{MZ0O?PlQqx=pTM3GClvGwn&jRav`65SFC8u+V;k=4A4L%{Yc?#Yl~M<1p zyJ1!I-$<*SPxL3}(kW9;7-QPG6Bn*7-Gu6<)~t4)<5iS;5kr~?+^4Ub1alPd-YHTE zJ2C}Xcuy$U+i%fI^W7}0=cjbjW;vpy*Eg%ghU+g?wqfmM$T0FR4ON!g1j!v~n$TBg zqPL^aW1PhB8%Yk=xFGI!kjnJmSmYJW$)Lp@XFFDf)BaF2g$KhIWO`{I$m zE$LD&TE`dRE?tL}T*|Dr`53c(tI{%ZyGfO3pYSL}O1rk>R79p5)#tutd6}_8tekjG z-~xQ5XzyDvM^;U8)+xfJjY?z98vd30laK6xC$GXdVUZEpRhsdRgW!0M-vi&W$jRmF zRHt~pPER_+r^LEIZGMO=%{(T;>BmLVEfNx}ka!~T-*VCpGT*1ztELs|t$!p=3fR5& zunO7yYK_x0mGC;;wRZTZQ{ds|G^rt3g9K6!na6}cfxEboI~vqI?bRBlt~^30WfzLe zCoS2=IoY6%6&<19mz`#n7%5*sYnct53l&=0?W^`c`KUESoH_a`ZQ~a8bxDYaS8aEf z&qVGu^RH0#qAvqjWCU*Qc1Of~Hi1R1dBi|}VZFA8-NTX&HAD6ntDy~8ZC|1NMSL^) zLt6>ExrH=Ddm1kM*}X-wDl z%YG?$+sp@`pbL6AHqxoScU7uAMB2 zQISbQ8X9#ze*~*v28p`n^isMCMb*j2=CL#rlo#Xk#IG3?T8o3AxgZ ziOq<|4%o6Qf;>8^Fz^$R(eOh=du23`;ap;3OIE`q!a|T?S9gdEtjaWER@*~ESkyn7 zTl0k~1eDsoZDZVBf5oNY&p6}sBV9L1iLLp|hCb2eO#O=w{lCz{>U3F-_(GqW)q5Gw z`}_mTuP5H=9yywXRJc&^7k?UqoLjgjg0IK2aozaJwj`;LGtrkL3kpV2#vA5gK((LSA+R{>hS$M#9sGO6KNo^K_a>XRjZDq zQnBvtLb9)JS>wThb(G;68cqqb+`U;0GeT_Mn~U6EIT)fy*%LwdLyoA?R~Z)+{^UXZQs7l%a?y_G7ep*u%C<~g+V@yi$lh%c;=cH`@NO1<$Uf4p5Whf z2IXvxitBS&?#|@AZ>ORb9v_-Jzb&stExcS}m>;_Cd#mVTMLtu)CZh56XidotDbd3* ztZzv+8t*bg*Z8r$lXGUx)@f7=^z~X!Mhr{yh_bg-SsNw!q^_ESLMSt zQ#k2$P_Bc=phT!pVb8)_tdgVRGH|PsvTZ2xkN4T2shhVz8yz;&Yw(iH;V<=9`C%b*Uc7wL=Ho1U~d7G|`}TS>K**e`fhSY4^lb*fb?j z5Shtp$*+`#1YolJ@;mQ zxkYjVUBL)%+M<7@wP~9)w}m#UCQLO3lT58I7cSVwTZ-9XBp|#ER^gbc_wszn%bfJo zCL$!{-C^kCJ9A=nqiz-5rp;PRpXcyGB)6woDPdM*|48=-z0(cZ^jrjQhE0VeuZ=wO z<1UB1UcR2jcLC3KXmC*~&leKWO>MIJog;9UOiJD6(VSS%jmZhV5qxQ8__SAW;{lO;>c5b;5Tfht{K+ zUBsAxHy7S(`~!&y+M)WrTvM@8ji3DAnEIV$eh)OS4Y|97@ zBg~@qcv@k>J$i@BHwm2L{EnYgu(z4rH2J(n^ey^7>6wN2y-rkEx$5NGGTb@TYVhD% zL4VZuOfT0UCW~cut9DPzs2@!-Kuq)cDME)SC;zKrjH130!)Pa>4cm^`v&&iFLwDG{ zrp+72BTT$~eN9{UMLvJ^{cs30WkmKwSTzR<5v|C;G{y?*RVBK1NMBWxTst-1X*|3L z8KXh(M!u~peIx0_vuxe>jgrx!@GOSS(Zy~yB-{Q|v~n;`4rmz(NYJ+tG!XiNJ-ro- z{n{qW5id<5JX-!_C);8l1!ZUn+QzS-)F;SYZoVg1=61C*b#x)j&8y!(cddbXCM@ z0^xWCgTdP`-%I3q34OJRwmpTO+UUn`wm#ZEzQg7BpiGcsIx0skJLOL<|L1>!mM4B2 z);UQ@YFi<+iyN?oxop}zoCK6+4>%QYTsg4+cJMS`TTVKJ_PH$@x}Jz^LhMWHIn_pH zVfJD0epRbHLiY_^m>PLNs3zjd2Y=?fk=fX9eTgKj!y1S&aXbKuNdo@YJP_m|tTbMn zIZ@97Xb>1NfFDb4q`E%`s58XYN^CCoMf=PPV$O&e5XghDGUY%kW3*@l-m*ROfl{TRt&tfN(WxfBhCBq|QJn07Jq9f*y3f^;gZrXy>5%`Fh#fo)ZCyS~Txr z_u#XOdjk6t9Lz|S*&odk`yIQ3TDU7px6aexrw< zh3b*Ugq#H%3XL)%Qi$b({*8m{JjKbY1%?ZvwoRthKN$7e4L1+?ueJf%c=CG z91)WA8fZ)whAzel#Vzz`@$#=er#Vgd{{PMhaP!}aYW^EfHPW@@#r{4T!MYe#^|bPE zf29XX*fj1Rj-jj2IO(Lf7dIi+J}3ZZ9ZLDXBLtt?OkoI2upmAwZ6G1ItsT8TU7qti zp>wFp81K}x!1kM82EBh6ocgPP0#mTKT|0*0Gv?u z-JU-PSbK1@JonORBg^b`8R`(?16HP|hve8*DQ|`D~-}_CJ$G zbt8*@|6QH|hXf|%+OnNf`^6mZme&lzhj{nu5G5+?ms_>_wOgevs5SK$`x;dV~I zsc71?W9we>IgbF<@Q)Yt#m3k(UC&`B!ijp6X&;h2836c`j(<(alp9NsQHTCAvweYkCdK=Bb_q9T%{yY?qjZ*`?P zZO{AN@7vSSAD-uAPqm!vxWct^aQV7^BEWt9QA+Oa^-y`^zLG-8yg2dkv!~eWY0(Ct z5JI*v4!=Srl~!oB5<=yt$GZ_%cJTFF5cIYN$KyO zqtL_7oskG3`wb4i%+=4V90IY!Z9X6xS*H5#y?=?mL0`>?p%seo2WrQJr4G;$)zQ%& zQ|EQI<~eVJ&d}LmK@m5*&{fkNDx4bS|0~A*joOA006@w5Cl}zqjj;dUUabGki39(| zIA~EGPH!hphgN%XhymkFpP`G5a^GRUs0L!afuAN!NPuD##uE!YtNXQcmLVczsxA{i zD0LFlok`Cv$@~8D<0<2_lXo(E2fB+5ARzKBT-ABMdtms(iP<`CbrdP|J`4-;f000}f`fKk{FrbbI`;ymkS+g%&#sR&IOVl$P zZI9>Q8M5hK;AFjpBR=QD^4;3V!y`2TaZFt z1apA6A`ABqnTJ4lGb@%u@%!`Rt|f-a3r>V}gbv@VdXERCYJ7NrCIbFVi$|8@zksE4 ztzvFiXE5H4Mf$Qsw6MQ0Lx*a(w7y;Nw{tY^Pb)}MPiERC(0)LD# zp2}gn;7(#7l07io_LnXrHp~)6g%wpdkjCI$QxVh{YZJjj3HpaV%5LMb)7h#kXZsUN zh1j$pSv@p#The}L@DF9kiDJmFghv9rfaip}AkyvpL70+^Xo3J^8d)+Aln`GcNc40_d zFb?bZ`l~VHB%``v3ZrE(*xr zdAz+T-@}%3_F#iN<vQR+ym08>7$HYN^U-5 z&|uDCm6-qN^Z7fnGa*@`0XZW6rVbR9RzV7(+JS3zS|V3#iiEq z)ll5FEMJrS(xTAG9AMwgbn5DYG|5CHlcTNyZVE$j*+BdbH4#^(sEfGfUQI=g=j@EQ z?oI6@yW|-c!W-nD$oMFOl%S*0s+Q28>SjI64~1@vzk1NiV`&On+=#+PAEZ(9kIQZB zICtQFUMbe7h*m=v=4#?o*__4?)Z*W6jehM%lGz1l~Aq|YH+R$lZIv?%;^JeArh98WpMq+zspv|s#iY}ULvi_p(u zkG8Wadv_%>m5S<%P=pYk*oqYU6SFVQW|E|}_Z9RbE%GZt;bkp@&t`0)QA!hP)6)#&)mbh-5mnowtAGz>&nc;?5o%qw=y70yF z3!A(@AT^0u$vLY1f=t(eeL$`YIUpoRk+wToh=>VtND#s|j{RPVIJ5Jp_kVe`7M3yD zKG7)iQ0Q9Ye3q}{9igtO?Khu=x4bAAC!YTc{O*&8SQS8zCVtygsBdawG&p}csLbp$ z*FiGI7=sY69=h!xn4b0s4lA>jjf{}@uqL&#MDR+G=0_qyWjPcO!*o0j&?2NCzTq-{ zv@6YA&HR0G4duuK?;i1qL}CvUv6}OEV9G$dSv?8moIDQnQAflX=uiwh<@oy-qB_|BjM!{B#OgjIzBrIdUYhIYnz~po8Pt1Os&1A zxGzxDYd)(A2nDA4VOSeoW$XM_Iazoh*@gWbCaeX9J(RrcW~o_B^Dg$X-9)!HTy18W_UfIvwOBc3EWo~Gg;^CZ$PuA zqDaV&I>jGw)xy44((A~wC9iWG6tz7Y;WC;fJ2wz{o@erP{}xk-jkbpS85qG*q`cOjtNWI3dPw~_xNUI;w(D3q>+O=q4fn}whc8Fu&QgF z2cxLjK>Nd#4+35jWr@9nL2m+c^}^EIjXiVq?4>8$M7!V!+@k8)cc7r&pWJgCK(}d9 z-p$Z$==JJ&e`24da@|)|SxEA0>tU@Q`(&RqLWG!pHiF`Zg)K^$)75R3Zv;@31 zsLEtMmh-1%%QW|N1$9tHNZke!W{)ak5?U;A1YwegCNgQ?^3apHphktBpm5*9$I^+? z3w89oV@ZD>?HDn9au67sZV7&Thi60Nr;jWE(`k+Ph7)ju>@K zQwSG0yrvRJjv=;TRGmMmh%5;*mVVvoTSsWF*DRPEnJNTT`Cb7mfoZO#o*OA(WOZgfni;l^|jOO|4>5KKWHyKdq=4 zQRo}A_Q$6w{=@)G#7f(!(<`rHm92Ew4xK2(ofpy zk`xX5uQW{2ieuwNpO{Ql%56934YY9gyx2~&FMY8cB;qnn_>09^$( z%~H17#9}TDt*g?v<)7DN1jO;0IFB~6XyRGQc|2I+i zo1b4;($#ujtyorX;vNz6w0+{?*thISdy}oK@j5l{7mDEwaJF5$b5of(ANLn->B?`* z*0ei8FCWHR!HJPYxXREzvya@$Wn5usG&(6OM)N9FdsP!2a1Q>Dx6V&FISW>yMYQe1 zI5#YaMeeRbeBGo$=)=65dgyy8QglEGBeL{EQnQMlLl)-}EE|=VzX=yXS}(DFSGUYH z6hFLLMJh+CVb08yfo;)F>e1A$@fsIR61)9zfaP+P1b3m%snoR%4BA2mE?mO<0|%Ak zS1i+q^inOOl?eDAq3n(d_loZJ!6oq|O*eZunhRvIp!y&Ay@RRWk+|m$Di5p$20%3I zwMeiOtXqvw#)o{!923!2eV$pIS6~c@Z7<^nV5JN_5=#vuO32N-79|u5?`45OpQX~@qy=b_O=b|aD5|1 z? zm7Aip;Mb*C`&g6`N5AER%WtPS=?(DMY6@5H`OX+HTki6<@Z#9u)HADKW!rRrq|%XSXH+*E+AsZKDJ=Q+7wmqkL(1vUagDPH#Bd zgjo6$dxS9Tb;nnvis+Zx;Z0avGs(x_U4W}09T{CU4@%`ZM4V_xm{w0q;yRAhcn*o5p?oFB z71V+bOM(*VaR8D6zAFJc(dq)!Ny+Plm?Nb0&S_F-`J9m7B`KjWlQ)zJHwwRMK@NL} z`-Ufw-`b0RgZ^74{02^l<-n_jQBn{)xaXBJy29m1W@>xCoi2KTZ-r#~Ds4}sa@dc^ zNl;+NM3A2NF}5#ln0Q=CQdf2K+w0oHO=;8W)}G$nsf#D(GLxI_?^;>eJ(FB?*gP-C zKUL}N2+32UOP^cn!x6{jPd*#oTkVo4r`HsO5GA){@}0wxs0PQ&Ph%lF>F!tBI~rdk>G{t|{Ctt=DUDEz9JA?_ao zTN9d#czg>GSa1u(8mp)X`8wU76*JRKGauIAj^FPg3aPVZU?0CJKH7Si%AK?2Hs?cl zQI~*eQ}3suDv=qk3|dxp`p}LTtzwJuWLLE}QP!IHsZKWL!IRTiOn7Cs#5`}Ba zD)=>t+;nQ3Q-T>@)?L_5nr)KKsj$BOXid);{Rzv^(X;od+1gd$m0RG4vjfBw)}>(e zpGdaHzu5@ih88?+ec<(M9riEnM^;K_a#9z|Dm%M2BH-E)XMi6GE%o&5!zC8D z=kCCwQbYN6K0SM^tVV#HLOymU>;Bxzt;spk#njWt1p+ohx^Kbc+JK&XboI5SJrvijO7bdTe9WNccxdo>r- zBa5$7d-L9*QvOCCKt2!R&WioO;EJs9W!4D2vxO9J?!4wbmKyc!m7Q6%k5_;wIvz3D zZ&KSWTX)ssmi-t-f;_#-9hmLnT$mb|+_6=Xq;rofrB;U>zX7}VBoMiF{46Fdm9N58 z#wDg^Z#XlBTXouoy_zxBQ^vNWzvn%3<3aXrGly*6;)PkYZFP?Y%1rbzjx8@@gI3H# zW#E#Y_7Yv$p5E7Eo=42(r9#UO zpR@7@R~8t|B+hr2;qk5>-^&%)$JD{%E$6@4mi&tv3txC9eaOIgdCMXlFeG!>FJ)Gp zPr*yuT9R4Rkzo^FT8-`ND{zYrp`+!}^)MC*fN{{@E5nSKqZSkT65JQKb7WxDVkdZh zvvceD`SS9klYMJ_TfWXU9YYf#r2;z;sdQQk4+Es~HBTWV@vXMgi}*lr@(hpFKF% zg|+L!-T2q8ybi5}xm7Wy1lkN-Zo8P>g4Xbdw%fySfd@N75Pmo$yzAc# z$HTAPoC@-^=0IkT^Zgub$?=crA<)-B)3z#uqK!L3=cppc7-;zHZ*oG^86?+esL2jK z0CRpg2hrwem@lMyYSAK4!pS%`?+RJ^hv^}|KX|C^Q4-> zOig#Be<0^~H)KI;&;@XCmw)R-;doHDNSt*YR9}iL%`uS2c$}S;$)E@mWO2r0)>tj` z#y~$AwMjrays1x2F3etk1qC<9~iCS1R|=Z1LB z+8`Z>7uOK$Op19%F(!x+fhyk_9KiQf^0`_b4*AN9%iVt(uJhkH2q;5?cXXJP?%uck zMw}KOj#cSSMl^KA@PHTh1(7tw-g?L&PK%1FN_1aWmkZ{hBMM&f&R+D#7bh`|(r`fP z+Dhp=5y#KJESWfu7HBy)S*CzC*N0IgLymQcVbJsF_PF5&B(8XK4-*GA3tb40e?#Zf zh4bfM+sxHDUFMd48$8K8sSB@f&rhEp+gD0ChiBv;e_6h&ZiNZSl`(_F$~M*=YMRRX zj`FE;s3xI2{oyUt%wB^@!H_ow=dMKRXi6JJ%Z1)p<@6h35ZwwAcn%4xQ$R=9`2 zFvmYc>0dl_vRCO3{`UfwnOyDQeB9wHrA!G!^Juv%uJXd%620 zt8BIR74h4#Up=yL0&c7x1~3`x^BKxkdNChcvBUmWE$lq#O}I;yf*u{D7Kd^EV2q@JZqisVubw56{Vy_t3@61E9kh&QUoQaCEbO!h#`@ zv5rqJmSz;@j6{Ib2u|WV66d@ZBq(js2>+DbytEzcyq4mLZkFf*V)oEc&|e6yAa$Av{Sehq*d@jk`F0HsO9i0DFETn+;KYXI19Q@lO&fLGpq^|yVkm4qm3E>(4aMAw< zS10^unQ(V_0|LbU&E5W2ievt>G8f{R5ynq;d|ay_aw3ja-+ou20qs{e#(&8})~ktc zqgQ#IC6V7ecx^&ITO|iR|DigC9NoX&*dGYr#5d}w?7$Woa_50P@UIMJnP|FSFQA?Z1!n>>DaPKeeiUgx>*NOIxM6)gC)$#9OOvSYO%KNsI^P_*UvO?9P zBrCw9fn=l4$F9pLjOV!qXj`eh@F)k+y|QP|_Tg`v?&g-g5v_?#IoKFi1^g(iLq!U52$U;Yoa{4x&^v&~PwJ5~B{wR<6Mid~Nhu&SWLW<86j)Esx?N|G=o zr8U__LLdbFs!)DCq3H$WaQ^)z=4|AAlUh_Sg6iKcksm_#hR-WZj3<=+slAve2QL~I z`-w`>hHw5D_?nQM^FAQgpwW+2-u|U^v3&|c9?ietpc}rjJSTt>cFOP`JKT z=N1FNEn8`aW|cxV90vrFzY{HzdDoZqP$jbX@`CIM$bBIAi1PHunVlLXL)7{5(G}0; z{p=Gp=jJRb_P;`B@KzI=ipqUV z>T!bLf3W!f)ncmuY`y${9D`p$l+fU5#rwATaq~3u?kmtC_S+1orQKYtA*Nl=qH1tC zXuI*4gP7Vevb^kYFY~afYLAL4DZxJFGIpNzbnb)mLE^@o;act9PMs}Mikjf9 zb93#KF2d$FO!$~O)_N#bN_DLI4cXJXr5%(qdKF`oyhlVTR<4?~?BBTfRB?x^jE(Y{ zdcUQ@h%YibmC9kz&1Jc6X#ei}N{yNUtRim)xhew)iFg6Qf4@9YpPAEgcKbT(gFQynC+| zBP8Od^C0A)gq?I~B{14Xm2N||m`d@74*g6&8_Oa!#sH_&@AtUXlJd>_Kw|Ad7e&~U z+7X5$`fH24vQg0cHu|l!Swy~BjEJ36h}yB+5a^6i+u%Wg0RLzsN$& z*5ld{U9NA?XR2hSc_D4`kWlDPBp5jj>M43zLCy|dCTksNiH*-N_a1}LoYei9Ili3h zRjr@2=+D~IhFfZ13q#6mKr>n*sT#r!LNM}6jSS^8?fvekT4c`Ec_wj4^hoi=W2a_C z#m1Zk2hU$hSi#;Hc#ZAuewn!HtyX$Vx>yFjy;A>~o#qC9x#z5V-@De?*?W@A^UO1oy*Hbg?`I}fS6hXQn1L7o0FbGvD(V9O zxG?MiAjHMy%r9Y>uy4n@ng&YPcK|>N0B`_ENdcs=X%fHzjQuVJaQKt)XTyF-l5%j6 za=@lwsU++UclcYt;ja_c;2@Rs$Npza!oD1muqoI9>jnlpfU)UKA}42(6d(y3=&vi* z{Kw_b2FAXUz}OU?1P3RsNS^6g~;gcxtVWnz5zSjz>{v^;2Uh#pF-I5hDj^I5BsBuJXB4+ z0RZ}pKgTUK-^ecj04qRE@tHxu_x(&KJGgRcWm~|q$+sC->z{D-#Dia+(hcInod(jG zolY@)J+WH795*+hpEIhoy6_S*7bm-xX>8 zS+dF0jZ)3`ZSgC31LAO6rFmEde|(LA9WWbz=GDHLE3y*>u;9021a=eqblham_Yd1H z>CQ@hMQs)Cxcj`2^>FW=_0q6oOvsxG1mlYrn8HInX(ObLFq6D~h|Af*Iwqfh1+{2) zb>Q7q*|`kn^a$rXJ-C5LsUrh)N_;mnB4A_heh)nc($Vn{l5((C{jko+HFRE>ysnWe znHYOhcL!X;ryHU{Uq4rO@W!VKWpYy4a1c|Psrsl+r-CK!*vIL4*HEAHmlps+Qop;D zjSOzB=ZluXR`q?eQuZv)VWyk=EB2RaDaXFQjd52d<+f=So%x%tCbF(2zv%PwMV>`jM@W5#Uo+$-xv*64qp3%Od9aZ5J8}Q(@ z3qxFCPC)WR+z5D~8=fCkx*X06XDvDt;-cGFXA!&wLzOC>9kvQ((PerQbYTu6>pD6u zOg$bl!Gx=UbIFM7!uw>}jIYZ1IXC?zIn3go<&w^U)x67xS!*=)5MF=l-lEeeshQT$ z!3#4?Dw4NO-)6dYF7 zic}R_)wWdCOHTE8uKFpWa&izv_-`%dOB?UrUACsE1Pz-9;?;|^mb=-^S@E}AUAngu z9FXh$>cie$-M{=_s^Ngxd7{+noA1_R>)gdl6*s{1=Dm$GoHcqI!<+To?SmQ1#-})I zD}iX8(vXGs857xDE%gv;ImxV&R6Hx1YdhVeA0~6La}U)kL~;7>mqyGw#p8S1`2hXl zbTKCuSBvQ6$FcBPXsQ;dQT)xqB3tfx>Tf4DHJfdxf(0kO^>Nj>e$_v} z8ry=~`2q|BAC7KU@+XTV9?Q9oSJw(E-=smVkE6;yriN*pSKM5GsuL{z4<4WHm>_?rY91l^I?!WMcwfGLmf>PKZCq!z{8Q?%5G z>|d5RpqiPqQY9U;K@~G-?Q#-l^OCGemJgX+UzCeTxBz9ZquW`MFt&72seZ2j)Vo}bi#soLSzB7*>fS53puTM>Lja827nPLp z^LQ(s8gsAI$(!f!&Emk*sxGRFD>ux=20~^v3_4t zI4rm=Hd{7zGRa)qp*oCD5bOw77TcQ!y33mKV}2Ki@U;PCoPek3ov!_rg2S0*@{eWE z<2UR)V&VK0iFf5xk5W^JNHiymS1Y?;($6nB9?YA?MoebwfXYs@+^ME~&mM>6;b;Wr zk4^7X>c51x+Y!jpW6(oonOs{}vn`n5m9uZ6T{^&F>&b@=mnG0cIk~^K zZZB9rRX4v{yFE3W>GT`L3w|u)cdiz@>5k-6Qr;a_H-mkxQ%-O>dU4mJ;$RprxKP+w zF?usOUGH@8XRgHv+8g>kA0`sYx52=RlqkO!)#Dba*f}M>nChi)FpP>h$l~2q$~|L~ z$QR`U246fIndSmL^FL;l@sloK1-5<#f4P6+|EQ2VOdaRNiEq8Jt*7Hd*}IVy92j2Q zXl{BC?(puD&!dYi??Rs~s;A?xd?jmVMvd_?OE-RM^If8awCq$Hiy0qOt1X9&k4;;a zcG>la^EZ;u+ngHPf(9+4yVav><@~Y_bWcJ6FMcaI@5kM05|8I!EhC)z z;v1az)SQe=GP76366x@R*+X5!_qVVY@Klxlx)<~0vlQg}$z%d`H=E&bzTt(W!|^%g zxRvYOjwRej6mRp1%ZCK&Te6Tl$@SvIyJQ<*zH$W1e}i|7+S1QIS)GnIyE3%cKLQ(P z)*}>7@DryL-01gDO|{`OOMMF6ZV^=J`@2e_L}2lLiL0RAvOdc|N+ZWjRaeCxUeJx# zT#;W)8;y4;^v9D9ax?DAiS8uhPi>jYVXRR8biA{-CIj9+`WR|&pK|Inf89#tQDZeG-cG}?aScaUD_H?>+vD3s`qg)E^t zX+l!wsDh9(=e95yZkjn_)RSyk8xRX!?in_7l%+qU$BScRg4n$GsHp6QM` zQ`R-TOyqPt!<~v^_k$yJ6dL7?_)&$1S2H{`RNQDY_O6LzlQLXBFA;g>70%F`?sUCu zEQ==3IP!^p>hb6Txz5VH8?oe`E}BqzAtQgL2eI7Zl-bw)7UCieac% z;{#bSjKPIcy~O6j_F6#X2c9LOzXcP~v3ng<@ z_T`C8o$2OE^_Xo+yFQ;C=S(uABq&La3s@FReNC zl)v=^X$FHNk*20Zee3ERkMUv7;Tmabr{N5IdV1Efse?HBQZym$<@h5yV6W43y`2J@ zIAw22&*Fz2!>RBNPh!gg>M^6(V;PmmW6Jcb0NK!BOp#hJ-Iv-Zq6xjS%4Xecr)PQe zclwE08pgR9AN3Xyg#(#ByJwRkmi-^*A23&!mf|iyD=RrOvUML8$Zwf6rooIecC&^k zUPc^Phb5hRdTH!5uG?e~ew+1<^6@WAFKo~3QwBXQ#MefBJ}E@@msXv$NM3ND!gO;> zr1VHZbeUKEgY#aV*?8P12)(h$7U}4=$iY+SnqhP8pu3P2K}5+DL5v%kZ3Vd0^PhH4AAEY`T91)xVl}F!b8}r6=VKk9;%|d0w_{Nd{rs=1pM% zt9}DWdEc}!F_Yb0(pJe?proI*vQzRYt{oue-0k^Pb^ao;>`Ssq*+TGr@UWW8>rp(8 zAP)QO>ateUbee9fF7N1j{9R^qF}%Bj^c`E9PtJo3hc-d)E1mGSMwbOIz4_2CLG|jj z5k6?wdTtb!qk*tC4H*dWSnw@ z7L^72=F;i3hSvq4J9dSSpH*DV;UnT13*ukAgWh2rUw=E(5!L~!w$-T?IVP*$pHl(UfI`ousG6Gacrp3$IY6(o>8;JzLAnJ7;r( zPD+%GnWW|m+KtIi-JJHNGJ2w%Ze3U?Ypkb z$pe#oWV2}v;k%y2y6Vc(3cfRPgfXqvlf@V{{a1+>yeCarvf>Bq78(4HL`BXNvXvop zbGbT~OtUP-q}xj6J5{eAR(4#PrQ`vBOil`bjztH%dWW~WTNovcAAv}7ogz*(nX^}p`l)O^AfiKcPfQz>yD zBmLR}Qx6>66`CEkB&D?xgvPm$inab!gG;b2*go|4nAcmPgNU#-pVmWce5k# zLG=8aiC^HyU7__H_vpv2iK7{7-d@J_+fn;849_!ILZf5kXx-G+zA={Lf#=Fb#N#)A z8an0v(#`1Z6f)j)`|e2d(hlM3no!&|_7%LyohNd=M(*KbR4gq_AHe&s`Om2N`X=23%iN@Ut+NH8b*DDr0d1EUxTYzBu3%i_%Fg%lyR}!G z*XoOc>Ige5$9T#us2)^_h`KL!Yuvi+19a7SbU~v-XOf)F5PZxKYW`(AUqAg_uZ2!% zKEXDyZ5d=p4#7k}bmB>6%GEph-$%L2!wMRxQWa`Q%U8lTFVc6aCM^&hT^j@1_e$?N zC$hz?@@W?M`YQqGe7sS`(hS*%q-YboxxhO8Gekth!#OPoQ`SI2Kr&MP06svPJO zIIHJ#V!stVI~ZQ??S@8Ie&t3$#P(>vvQ*CxfMEjWOh+wQ1 z+2xMmSgMdf$?>2uch&>y_y)rAPsW!x6<65kEYzqkGpXP6`^#dbCYHlqN{SL!w839# z&GCXot5r3|Z&SNd&+s3R{&n?Bm6$-}X3UJ+?&mEl$3Vyo#Z#Y;`z2uX!9=!}TLoPi zc;G&lpYPMaJNYj)^@5A2x@o^Ltg0Y$t9}=nj#{b7B_Em|J(%w&A&M%dV+JMUaGm94}sXfgyj zndqf;Ik#*}7*VD_wo=*=M1&#fO072-3OfPkG`%n-_5lqByPmCEvJ!O((U&BTeb2^w3T~M#YIE6O2MQ`jY(Qb{s@Hf# z--6+vajs1Tsuy0>HzFRj7Ua$%JI!RSF{3Nn6rka_u3G9YI9`_8JEaGj6MOK~J0Pu> z#Mij3RJRx81hNcijd!fHR!;K*i%sy_Lk3%I_dRd+1gZ<%gu^u%D3_fIe2*OtP7gJ9 z21I(>-!Hjs?C)%Oz&b2(>o~ARE$_o%DB~4p~|&RYHIR+Yl53UFTY3(7`I%AhEF z`^~IZ4`7zk%;e+rRi>^yULnB&<>60oR6vSIn58V(kLZQ}8({zHg7aK+DMiF|hqbao z7CT^&o9g*WHAcb=gqzlZEaGy@mgoQGZTczRpGqR^2q0p@Xf$*6=}JO%EY2L+IXBVM zT4o^ra`bkBwFtoS*B$B4bz|V*+=iR-jX~qHxUNWQo&kM4ZCze~`tQ3ELSjIIGC}|u z5s(#Lq3R#(Ti~zHIr&>Pu!l|07BBD zVYW%vkqhtUW69nZsxer#&8EGs6JbXuv3-9Cs}Vl{=*7qWLSK!J)t)d}sBaoLo?9n-k^_dOe^WFYztW;&X zlcCaS$cq&aA1EF>)>41ua~x#=7J9?09|K^dv75ui(<`Ps`@8JQHURxdL+)EPn#AF9 z*A9yg)_+vPw|JOFnWxz3g#WQ<>cND?UrXA&0D572pf&X2eq`f4zqBy_Kt-W3xiO!W#_g#TIrIXAR!5%yf2#%;#`0emo&Rgn@jt@#UPsgGpVEF>A+vHfE7q6c zi!%I=B(I$Yb^;zMwzX3 z3X}ASWucUK>$TE-dJyV>!(ubf9`xqmJVvq7M~JZ`v< zpH?#!$AEDN%a*+Ir)O#Bw}7EDEiJ|yeW3|uBX#dpnnThisPdK;v2FWzlji>$S}mcm zU<|r~o}NUWuWt9FigiFb#LJdVP}fBwxRpS><0(9H^K8MA|LGE=sW_-?u@!@^xjxe7 z+jf}6n8+kIAdn5Ax~}HLAq#Jy=Y9pcZ|@O)e^BV35`K`{h1luQA6*QcalJuYonPpA z7BlKJ828W3E`ZC-5bCRt!Gt)tyB|DUL_T8s7Et&VBB6d4nn!oqlM%#ddC;@&duitm zk7*YGrR{f@U8li1-@@+IAh*EAbz{1WD#3F8@kTTQ007JVKXUA2IM_m zBAqqhEDUoApOJHCr#4(@xxuJ;+rSt7KSWt+@A23J<5RJPnURqHcit9!P20HNV{_=licP zV`Xa_$c+MG&So~M~&lz<6 z9bE{^;Ptbq$c?a%`|CAf=-2XJja-B0uJC}I6#uSi4*xed-~W&47XR(rAq|4_xQ zY(u?y%xVoab~=hlNJb8}!}Jv35H6NdpUskjV~j7asr#2N*2L))?i03;WsBu^Cul3D*^0|gZqdlU zl>&4g-TDXa@?yDDtET4~5B1;1JoMn6+3x@HmvstW9wR+NailUk@(Hi1I}X=zDap#@ zD>}{=3%J8Pfbx&&8;JpmkxRoUgz1wQ0$LM$0bcwMxfiw9$Z}It&Y&Y);x{#PH{QcAT+Gkn3zP?-_PJBxPUA|ynwhT zN-KI#pC`nPX79^~YgmRj1D{_j?R%nt_c0wOQ~p%lZ&0lq_uYNTAoDpfdD?7!o496jhT<|Vo$seg_c>TU z!{D3AcW0Eal%b@w?+#1&=)2KT^l^z)`02JW)#|+pU%mN@w!rR8Z7gHszQb9{CoOZT zbhn!$OWq|rI{yXHo(srwg>7o3b@##H}ICRa!UsfWv zuLuRiU8-B|m&lO&SzXG;!f1mROMxA1<$f<2gt>fx`|h0O)&xkCkPsjl+Id3s^#{DY zr@qsVk7j^tkyF{zo12?Dso2i+*}UD6j7NIQ z6bFr|RZWLbnQZ95@3|qNr^`ND*FR+>e;QLaXrsOt1(ZK1g3Z|A&3rL4 z@UN#GHnpv7PB7lI@wl@xXyb16zNf#l;T}H88*RSKX6rIN={M@_Red@P8BnD0w5qynywf-$83Iz@CzhLn^wcp zFz#Spj798bo)$qTRyzh)?I9q-bE3B4K-XCISrXGlhU< z$2HswmNx1Rv1o?P-VI#4y!J|}4-xf!)ZXl&6#9f&&J&q4|>X8B4kDi?;NPRQK3J)ityG)g&o{W=k$ZSjC`KA~_!((_yGh+7H!tK=!G%cr@Y(u>=v2CjcJ zF1gejPtW5=7SFz2W+NZ)BDFA`(N5$esLxpNU0P9iVZ-v`Hsc))Rn9`FpZEHgI&PU$ zJk5s!2rK)JP4~kp&##VEyFsYtQpXOLRr>-=rb)3LvNMk(HA(AT@-H_oHd>U}^)m@o zbHhhGC-rk>pJ#Fo>&!-&e0ezd#)6m{8U4hSJTI?2ZRLNsBBSL#`U{9~S48=Yk)=N) zK10n7c9d*bzX$gj6_K_S)soNNFL&=P=rS96;JUNANE7(gWOIcHdrKSW)Y$|EtCe1L zJmhoZp>*%VB3hXp=E7b3s7=JxiMxQcfyr zxGPy7(iBOFdBz2w@%+4FZS9+pK5uBoTOUS_Z+>6IX?r|O{IgUx`saiOT}G9|$sHG( zEHnIW5#v!-{@?24XP1#b+Y0=&%p!eo!tgscQI^R+B3n6r%){Ii&I`8&d7a3R39-6iT*)xNv z7X7?g)t$WUdv8GRHF!8HO80LsDtV&E>SC86pYX(73Oe9N8-q192*%k8GJ-iloisK|KKLsDS5eL%5R=hBoEK_69if@JP=%u-V z)-%v>7ZeTs&uyIb>(_4v3q(1;ch0p)Yiv|0*{lX?Uym93F$P5zRM~=#?rY)7GC(V( zZfOxf>oQMTc~9Z5$u=LCl>k$YI4flLc~4M~S`mM*^xIPqmo)J@b2$b9+4y@-!T~v& z1$fVdr~T6KM=hQFX`@wMH>Ptxyd6p^ZcjMTuX=z{q?AC_)I@nsH_bh-Bj52K@1*wA z5{&rC5$Lpo@-Np5V2fO3R}t5jbh?ZG>`tRBIYuWY=3X8z4}9BQ$xM;OoA0{7c*fyS z;-_l8@oyvV%8_xa>ZGBewJgS}**yDb%DkG{Zrbz!4yoRnnmL0{%H1#|>zPqQe*#fp zc}keT7yP3R3&)76AQ#*itDo~o{t9A>X&dHkR%Hk0W|3zTCU{|sq-Cf}$uXG`9X;`C zh<@>D5HjgwPv5q`!fPsufl0#>7X!mxw5$_%Ux8X zw!Hiq{K5LZzh%P>iiyL(pp`-)qw!^_vzNbN=1NYLGo|*l1C{U+eL~Hs)B}~WR^kj8 zTW~U;bMdjChyObB1os;dW1%dQ;AD*>D6tW3J3S%Ek*snr>X9F$*J>Syjj9yX(_Anl z&rQK*z-DITr;dCqUAxaQYooywJLueF@XR>4F!E()liT$CB(cA;ty7N>oX5!8d~d%# z?eL>XXa(EYxN77F*ezKqx}WkC(ARzK{I@DXJ9VY=TfU8(d|9>N48%fQnuG~6PqYc~ zb7$REGN4|NSXaSWFZQV;8TxV3yEII+X+#DHE_!Km&%nR)8??c5Ua(KXuL$}Wk9?*3$ z`<9E{d-1U@|2qV_JhWq(M2)0VcP;-B>Qq@Zwy;2#cPezdLjE*@AtnuXI(_){1wT)j z)x)OapCk0YH{va?F*)N90YkA3X5^e@EDwCiTV%6y2Cmw=--B zS7~5b{gqnPDz~#DtNl$no8*OKnwQqaJ-4LowBKF1GbVW#KUnKjC*@8dVsH4ozTZbB zCPi(hM8Ay--($k~CGyp))QvTKq|9hHkp6_M_Sn>gn!}U4n06=1rk9^u{>Yk0LPaPD zPZ2J!JE+AcUT$*Hti7S60cRPe1e+nIQ35;HliMzEJznpRkg3Oe^IprodVM^Z1bd=v z73BpbPAEtFAWn}jD}jU-&--Ec*@KTM9yp+Vqy$@~h0O5YNrcLW$VP zl2RPP!D(G9JYydjTe^hrGyc2Z+HQ35lfYp{!ERuaX(A?Wi_Nr zq!z%i4O`OK(WKV!n?T6_XqO6K9~D>HmP=B$J);^fu_%N7UecKG|Me;gW|c0i$QU)w zGvBK)xfke#HVWWktX52J`3xmTlV`^|gj=|+9r9$Az@(2KdLf>f=u4EUy<{kTx2U#b z{!G7-PVqwx;cteW9KM$YBHvN{gGC8q#tVJ~%Ra>ZPxP3>#;*q7zq7PW^u9npdH?(I z;tQ9@p##)WFyNMckvJjelhWk)n2B8$S#5(w;(A^6nE8m&AeUO>Kn^X){OruOBi;E% z7j$Ro?D)V*TpU6>Y_*h$6kT%-Bcq-DoS@>Fo{PQFjGZl(CwyzrnbC<# z%nYER#h$nQ*|On8;;|>mm0g>xRcf|fOnNW*mDH}?`jQEFL-z@k_fwT@g`Q+oD+kv{ zf=m~wYAYiD86w%$gS%0Z`?sv8SB@!7J};&7f<7E?0V0FBXYYr9WS`lU7h#C5r}GL) zniM5#={8rx*AbBYJCGVAODSw4LiY73FA*!LJL4~VL|+D$+zU6fE}RU1y+wQ}QT}CG z@SK4~i&?O6kJ7F1&B6PH6WWB2FG93&cu0}t2@KJh?5V=F{^z^-peKY4vJ&rXKdS4` z*wIh@aA6D?!-UD@4OL7v(Pvm4R!D`h+@;_2>OUB1+{%dc&NW| z%#Dv%UfPN9h#+Y*B`%kwoRV=AD7&bY%OtZ)KH7J-Yj__O-$@)TdXRBi?SJz?Bb57p zn|%1a1lHtK7sR~pX;c^017PMjN3WIzW_*yl;Ew@;yCHRfl@^Te4Iuy@|3 zD-rj)V?~q}hc0;PTkXW7q6MQrx(7b&)fn?{uXtblNk3PpF_`U5V0A+8!q9kY%z^N@ z|E7<~e72v2Njbb_Jm>0!>Df}=@CXe(sS+*fu5R`-qu}W3D;mdeo;Tz!?u><|ASl1$6)9{x0a8ZMg$>NR)&b>5Q?Efs8z-J3t+$>n zLMUxLzDFu}kF+&-kN9jcjJFjtgRp`h&)PM_-R`GGvnYAq9IW4j2Uc5H*bwi3cyJvQ zDLA`a+0t}zJOd0?FlxNS_pa><^53V3{)k{T8o*9lrqDTBWDC^MN8*&x$gO47wnKer z+`>0f{DS6bkXCz!L1op`7|qR@3fmq(fs-RtFBvi#^(-?qrgOLOe(wTi-$L>NzEoc2 z)A6nhx|1=|<3@mUl9sXTsh~4YJ*CD=@dPA-NnmoPevQwgGh{xPvY)x+Rc*lP5u}NK zHtn`snDy{MYgS~ds7OqGIC5g+@t%D-==AntJ9W@UcHfXuwrm+fb9ZiZ6A0vkao#go zT>*LgXt7`z+Z%2&v0%i+D(yE56)%(_B@G}@#Im?`sChZQw1j3|nvRBoY#i*_^o zW*NfS|F8=nLkFxUW^oLe<6#N)FI+xz#4a8AFlqC-D>PX6aJs4Ll0-(Ep1=~xL|U|9 z#KyQV*uD6UHO4-WjD-FpQweT?aT#@21?W)1mbGa@n!Ft^Qkv7qal<{@ayg-=bkTHw zeKINFyN6(12`884&-@=5-(|*!SW-Ojbu!o2r;;cGRDEKn^-S2S1Dw>T(_QAGbH4DW zHt)?*;ZVf!ACQ}U13grW@p*I1kM}W9Xm5S`17(0oos*-ni5z=}j~y@3YbWb5XktDtR@=7wH-{R}Y0G(ve*c@76mZ1US=W3@I7ddTf894g(L3D6nE(eBJl&4tj9 zK0`R>G)AATiQfs?`McC%n&sr>!)q);0$Vn-oJ?o;kCo>KmvcLI3(-d%7tLRR5w-1K zi);yNeH*aTe^2%G2R#fRELrmMb1oX@Ku08w^yBZokI};B!Tv?uC;w}PLCG?3DH7&L zIYT4a`x>>8_5m|6v5?s>RP|~97;~;^u0Vn_{p0hXgL~G+)*$10g`j}**`V=+y+qVx zt@8q(+Vm$@4_r^YuS)7c!u{a5bsgWEZqLR_{>PtEM+nf>u5;^xx$vQ%OB<2`sdbNBr4Vlx{p1(4$DV&g!5CNHME)V3Z9 zm28W@$Lc|FPGW%-14`xwn!MOp_x^m{Va8|eH)C3u6}PG@5CKl zgkP)529g1TpV}-%txI{>yGce?PSjgcw*)>|sQ1kVCy3yszHNcsYX8VMJ;-hu4>O*u zNQ?J6=Ob7~XAhI!XNix2-8rNi!4BI8a`;|sjs1qVf8F$V4ER36a!&GN2K$%fAs@+q z;}|68F$wYPsJXxb+j4vi&Yh{+k%rVsJ5~6!#%j_WBVT-a_KiGN>hpbS{S>w`Y+tLb z9m*54{rGdp+YiN`UIK!F_jWT)Ha7*zUX0xIgqYoXBnYsh?9Vpzr0c_+zuuCeGWx|{MUuLdg;pW)g$?OfQtrznoWU!}8>ka@x&NMcjnb&?d@;T#~Dxv z+nAaH@L9jfn;6l)U2em@1_*L2e3tt2tu1zqg0>#lhBS}p!yj+v4}X@Z^!MSw=)Xr} zmnC9-0NsCz(??>>Hl$g~IB$ixuO9F@PGj$fTZF7Y`8sonxs_Y#qPbu!L+)l90e`yv zuFYm4YEBFc`_Q|xCHohrk`V4`(V8azrKz{^P#p2)y?^m!GoHu;U_O? zby6^uxnHkf^*;>+*(`i5Ob~M0FRR$m1t_g6Guuymm?TVy=ub%<)+&POercBO%IAde z{*`oeSsuB=<#+hZWNbumS^zFkcE+Rsl>cBTZsk=rI-7-Na6ZucdqPdl-P>pUM+7** zRs8hK0?b?~mKKaY2vv^bocp(r`H28n19~OgiZ_1cWN?P&Xyq8Yjxu4PGTskqI+V+j zV9KY`q$har8Gyy8>nu04+V{fO8ZreBEAfD)g9_4m9tU^Av8HM4^q-Rs9PJ+FhazrqOU{}53LfLqK* zoS(!4jej8M1Hpj{S`26PW9sS33%@^Ryq4K^%oBYwymzn%9#|tv+-riXc*FwoH@xjs zAapJ5n6IW`-mS1ai30C$%$ZEB%O&lL>8yRGllvG7;XSEiLd>7a{fY-ZcXhe>dm9~I zyM$AxAFW39{0kHl7-mfj!>m78-gS>6I;L49Kqu*?O?1hYoJ$3cudD|u#UKSUn z#B;GpwUWg-7Vxk+ky1*-L2LLJ;|!_+!P8O-uF!s2AKdxuPy`5*tSKFqMufk(hA?U* zPOQET&ARY#P@e}2NaJP6P)cZ2B1PX>apAdrxc}3BR-C<*&;$1`}-wvZk4* z=%c_X!wS!roUrJ2EoV85-j%V!UX~uh-IVqT1xHol{$na&lyh%&a*Hp&G8a92(kfrZ zXCZoKZwWrr?}PgtbVk6&@v+c&*>96zQYYpzgD&OcW$jqQ&D00*H`lYXf8y7!ZFK#{ z&QY4Sx`3YM{Is?U{#l+aH>Xo%2+tCefA)tS{~K4A{|8U9e~{w94Z(pQ;vjlE8oPEL O08mrXR;*O8iTEEsS7Ffr diff --git a/docs/README-unnamed-chunk-26-1.png b/docs/README-unnamed-chunk-26-1.png index c2066e3faea7fa4fd5d14b5b12b00da2451f5fa2..da882f9c163d49b17c1a1a6294d907c59749ff2b 100644 GIT binary patch literal 13586 zcmeIZcT^O?_b-Ue49pORoRc6Zpdd&_qJSttB}yDrK!PwRIogs`5Kw}WqvSA%1SK^f zA|go?kT?jE!$^+9YyAG+@4Vf=cF)pP7=gqC|193nGuDEC$;cGR$Q;NN70DFgXB?SB4g9T0=5P{l zayTeDz*n3i93BT>qYnSj4*w7*asO@yhdB6jh=ZRs4lt$$zK&x>MB)_5;v5uDV&J%c zcU)Wze8$zl&(XNinz+&9oEtZ)s;X)n$Z8xEYvSO`;lJW*;A0d%YewPcan14Zhrqy& z1rRx=%Kt{$-rh7%**f(t@ojP^wQcB^tP+n3`n69vRcPm=gVYmu+)k|2~ zj`Dk>5Otn}?$6_6rH&!d$+G&bUuy@-api87nf6QPJKqV31dw5_OYNK@d#r$>;b%us zU}ykx6a{z+5ekt#QOAU%?`})6M$^S)Y+A7t>3A8Cn9{{U7z+D;(90)KV`S4uMtey()f-GF z(CStgdJ1d|)4`(1fdKRg9I?&F3kuMnmj5NLhhW8?;MK4aQw;2_k0BeF1Ob3SFv(Li zVvq(VVM3L1z$C)uc^uIYnGVKW8qQ0PAMfCchqNlycy7tGid1h5nVlcErV zv@j_H511>*4xXap$Ge@N)v)7mK^Q>jBTNEh-x|c2%fX59V}^C$Xvwqs^J`Sx^(`ox zC9?m?wc*BUebqohRYiz@1?LIpqQK5>Jum2j@shf!Oyr1qC=sDHk z66XJee(e_iaM1R5dbFdnGgkDZfLKx73`PC=obuj#kIDLgq%oua2!(fgI@dcqplWZ1 zE&N2(YJ`g_D$YwINBZt3jPVoMe?S-ol=4@JFIvjcomPHKZqkSSYo?I%DH{;-HIMFzMr0 zGSqq!m;M{-3?>_(S~)g-cz`wd?V>cSti${#(iKO<3=;?>p$8g#WTo*7u=47Ogxyys zpx?l-P-t2*9g{-=>e%s9exe0pTRkbdGs7^z&Xss=myVSDiHcQHj)6^Zsd9zQ9s$ z|BB3~py-c2sP_b`0tmk|XYBq`uY&5#Vr;vLx~sv*y$x)Na=Wi=?qgu+M6-A%dBJSP z*zuD&h(EfTI!%HmXs6@rJ?%T7xTypYoFw`C?@GX`HV=iKmsHlDw7LixR710=Wz4wUi2{~t5Qr!GD=9!sXhx0k{nPLNG-1r}^?!)# zgz@?zRpg02M^vK@x80i)$a6J4ufo~hIRnhyRru=Be@fenAO7S4Z0ANMTwetoj5_{Y zlvg?FQS5kt%(?G2v)yy4-359MkKba&aCsDR8^+=nl54W0y?(MrRR7~E5tXKpT9d`M z%D~Rr{v2_^qVH0+oi5Cv^`M7IO;)I$z6KW6&WY%;m8$Dxb)0M*3G+RvEaU*`E2hqz zS+?wPRW>&%^R$5Erh25)PIi!kQmGi&qx;2_az9Gw5bmhh^~_E1 z!G(WQHSR+D$22rERC;23wdBJ@PY7I~U4ieb#J-ZkAs)3Ld{#VRWrz@;zfui1A{pHdYb|O4Ip$ z&E)wbA)}OYF>cx+e(Y#!_wUNx>JNG-ntTMC{IyVPbl^nFURK@N^KcfHmqTA)Ad=en zejNu_Y(zekYG?nZ=uCc*fn86XK%+zQ6j3SE4_Et_@`$^Gyx*M!8X3tj3Q&6AK`~+21r>{EQF0Hhwj=xxCn63&q%@p9`>{8 zD8VjC5YGd5Vpx70-2hr$SQ^Eh+tM>(0Jn7%6);!luOvXOz%PzUn_gTAgK>C%Tug2) zlT`9408TjOKftcRY^a#@Spxre6kP-gh<~U@EieCrXYu2}l3FIJojEO8glyw{U5=&h-U^S{0}oOh)a_0GT4~#;p%c z@lxygOgR+@CH7KjT1BZyNhNFP0+v}bh zKJmJC6va`I&?23fax^;KBg|ukdT8c`x(YM zMBV69n?e4K40!$CmY6=+XisfXNmquEO~*w+IqHO0ZW5d6#?3!|n{hQ#wzTHM`)HDrlk?f%;;fM@#JNmH#r>tX0{RzeA&Zhs3Yk}jAN zwwDrqXBe0>`6_jtddz~dnkq%7n>GGX#+Ma#q4!iK@mE!9stKib{ z6Y&(lz3jWiD`?zrulU{oI;1C{a^3y$a6mSLxoq_kt0`fgW;K=G^^@1)@9_%Txs^mEo3# zyO480cBHDtCO;8unuB&;lepQz@Vo2maPz6SBg6_&cjPhMJsBgjPwXdS~Vic|V8Y&BYv%;f~eSDMJc>$O+4Dty%5-n|q%kpL% zC{rh3?#iW131aF#&e4bPV*z*an}*Jv{xH^_Qb5OnSUEP~ ziebM(xP^TXlFQ=T9B>pXAt&<7l|*%X+2STDjYoSom+UTP_i2@nAV);3CSmz?3q85s z0)wv-enF^<(U&FmrGL5H-O4NdO;#vn+udYD28rbpHNdA%0BzM9_=yY%-rfIGKzZ;_ z63?e=bIt<9z^Q2rT*Hj%QQ^=A+;aNbZxBeZjpz;A8=ea+zN7U{Jcw-ul?JHMj0;5C z$zmL_4fRcs2Zu7D_oSP85|*iht+v%jFivygirP;@Ubox0U>q{>KD;43zcaI!Jr?!h*15HJYmb&^_cp+^_Iu zU&Sxa-|_GF^~I(Rb>5o5?7UYeaHgA;-20zq%A!&Ht+_Q4?r?Xz(t=u97_4s&KMP}m zB;>AJ_9cI`h!5?J&tGxk6qWFPss%Smb(qB_^pVqpeO!eIZn`=`!Ff&g3bu~|blF5( zz0ayn>9hXza}ySVgBzGDRabpl=bur-!qtL&=u80Xtr1C@&CqC*z6)&dl(Oyj2zQDH zSNHp40yDj04m!pzjOy#;8ET%CBGax7jvr znIrEQTfzOPh+zWUlI}l&yvFthxSx5c4!$%$83W3ve`MYf-#jYp&=ACXV(JVsRh!Lr zLppqnft@4XoLn=)!o&BBAQ$b_nIB^x}2etVq5 zCpTZDcJ&_>v*)J-qB`uEI6<*IJ&`Nezx=rLp3C@KUA3WaI%!bi853gm<}s`y_(4%_ znNKH(1sYWy$*Zm&K*+9)f*-bzx-puw?pOlDax~OQTs*H&2voS8I|tp$XQH#ck4RFd z0d(%+#lY}Qjo2SmC~8P|3RHWGP113QV;f6xs9e)o|fZktzprLbIl>e@N9`xdIL*?8f#9-upR1S z-Mw>#;7hX_9o{r_uyF~#*xsJdVzB-j&}r2Q;0D}_(PofGy)lpve#u}CoI*6k ziQ^ZpilZC*Pz_{<)PWuXorHa|6(4$rP(PNBv={k$rE(qCKf>!f-I-BqD`sQYp(1%AB9>-iSC|kDvh=&Ahb~ZB)y{9>kAE%I14Q1;jEEesG zT_rTOKX7+ZUq~B{xOJKcu$SKtLUVs;(GT$sIr{1?z&D+iP2S z5JhTx4#%$^G%%|8J-mADVxT#=BIkkQo)8r0K)6laARLiGWIGE~1&LQMYYa&=K&|}# zhCsmbT{zve(T>|^q9MVs;LOF=H=2at(DTy2sfOCmm4qEDX@e$ixFc$-_pXPZ%`e#? zd5JzIfMX(FdsL|ux=>R|4xWFeL%2%AE2{TV8tsyoC<*Ksd0p_OeRto1L4H=W&M3(( z^LSKZ529&*ox6!EQQ<~_>QPg*0|K}n`+idQ$6Uh%*7iyozG6wS%hPYsJPLOUdTj{m zAxLy!x_@e~-7ayva<;4cTWchAGGU*ZW!Npe=CvPuY`F3 zj%vrC#vJjv%D2BeY4EKDRIvRBGu;%9gKX}Kb5sDadtB};V_ygCAKq%Tom%+RSlW!E1*oOV}#jA)O=;fjET!?f>oC~o^fcXg}iZRwnv4JoCo+XAgqm(f! zPoc%c40kZ7!(842%^lw2HuUK`J)a(S$T9^^olE*0k!bq>8B47f84C!w_% zd?ScUjKe7J0YR3Usm@j7__8iBS<=_Ul;(4VdMxnG?ss|u=8*sDDJyWKyf6FW6@n&?dn%&-EZB4Rglq+9ID}y zOy#BP2_heA9)LlJ>7@v6JGj61dzZ(Y*mP}_b2$A5yDks?NQ#7o4BCevi zkgXDaVqTha(kTA$*Hz2zZgTXp{Fi9%j~v&0f~Z{UwRl*-Nq*uL-!)+A>+go$Ew9}3 zW`xG(ih^HA+54Kk0=U7j9@C+{9n6DySPo;suiB{t@89vwcwjGvt1mq_1{j*Q7}NfPU#py$K=2klK94kgH5OB&I;Z<^M^&~6UM=c8Ei8SX zdU^lpKJSYQ*9qRL*#)xi-$E-y&oDF>x_Oqb}mg?0%*R- zX1=2GD`xv%MCmUps#1mIscm_L&>zr(?Trs8W75y#SH#*JfUrYuD>Qki@)K)(j9%Zo zBAT$4Y(Swt=V6yi(`V6592>baxGXn~dZm8y^3 z5cQqq+;zwmApZ9%tg|(?621Jfl-?M!Lwb|SerVJ|Mx<+ubsqGA3? z4J&@V6zTBT4DY1rcmB_QW<7T28f8xiB7C;*t* zVd}#DsrPXgs}UK+S=RlcciSzS#5`??Jky>URSs!w!tdwK8gUV+C`Kr1IaxcMN=hP$iHDO&)SPYiYyH{}Kj zd@4Ib*N*qzk_K{v@?`K%VqR##XTRkLtsIOZO}}w1I8BUV>+Y?j4onVwrMj=HCCmPT z1upA=)KXTdSfI~0e*A%NJPST|X~~kH5_EtxmG(7uE8y;NcedIi%-wQ47xSDwAls33 zI3(es;x=r>dVsdfZ>W@Ql$8G9f@60|<+#R?RO~h2XM`epj_YQ+T3pj ze7@6Nz_%sVB~(1khRFA~OrO&Mb7ZLi@oM+-%UsAd&AYDR0uD~;JeA>DuuWDqjaqpnQ4=hy zbwvgK6vJgUdxhq*4#%KGq|K>ZKUVm=R-_Zjf!dHc#fZ!}e}LX~Q4K2{T2`g{TpG+x zHMno&{fT-Eo|#ZDpUu-2l?8SVT&$6!v%%cX*r9{s)wL}J-?dZx#HmniWCI;2m}VfK z3tc;}^A{+Yx%CRtBn{H=hA@g-5a#`}%iNl|QOb>pYr^kGBzQWAj8dH7i(>5PSnPBu zS-Wg{U-3LY-id}!z^f_hC!(nst-Ru1KxLB$RqBgG+hQZDy!Pp`0ua(@R~{n~+DM=v zRmn+n)FFhsKOGQo z%jXLroa!)}B=vS&Dbd?=*e2~pz1Z4{E+Jcj0&#(Zx$e^T!hi=@O;giFVqQJm+o}?8 z%QU0wRc3TC0Dp>^#=`Bc;1Dyam^VAU{?fPOO1UB+m)dQZsd72c{r!Grk{Hq6FURm} zhmIJVyc+w_Jx9i2(R?K^_6H((K#({Py*O8m3QQOBIo>*O|CO6}=eDoJzSK4u8B^(6Dk*IwXu_S+D0pe z-aLTeXCsXKQQah)SLFj1*Z4gr?!Br2(-&!Dm&GXTy2bD#(lK@bl`F4~YEu{Pxml0; z)iS9&r*TE|Q}w_3SNsv>3n?6jD-FO>kkgKxebVQqJ4m|Mn%(1D2saf?=gA#e8k-kT za*LCfxFxpaVJ#gfqc~D_$9c5^zh`ot5L#|;ZyUn9oJA$u^MSD^rLWV2p+5GPaUKV9 zaJg=|J!g1<`PUz6S03ps==P9FBWb1V$-u*EHE`$kmQ1CJ$~6O$%TIt3C7m*>meaHP zL5A_U3Biqe-{~l+)&*DOZu2fl1UI|v!7?)H`mC#3Hv2AS*r(y~3z{_`c7L?|2|ZB4 z<(lEeV%0-g&XHP^!(i@VIUq0m_Yb->AVM7Jb}(A}g#JZ6JPSW27B^9?to|%@xpP~D z>nfJdgt(6IZU{0WphK39%we-vynzkmVZQYNh&8<()4E_BmGpvU2e5?~Kg^&|l58tgj;ZwAz|k`*LQ!B}@pQD={XHypqXq6)(M;a#V+2%Aj~gf~XQ zinW^F*Y}{fl;^PV#zyUJB%6;T$|`*43^DN8{0h4{)@M%pGK&kyv(m>8xuH;`p5dK zA*M+bYD^GU(cCRy5kw(~X9MTzx~qo!BcZ0YaF=`YYf5!GXNV>`geES+EPEMAlpTb} zen+V;HuE9T>o5b~nSIA5owrG;!Qd8+?6;4nQOkaw>4!3E5-UP~pd(83&#m;#3&obs|kaU4;0d(!*M~wV8 zgP3NGQDA9T$qH%66_}mr!ILC}AE5Hh21;bAbWq~>Up9^PJcw0O*cO#7!Irgfv!O~e z1}FCLzcET54o8xe(`x|A`bx66bGGnt zF4OLfQ$B}{_p#PUTYkLuwI|3y`g@!Kvf!4_#>sdS=yA~*@5T}cLY(LivhwEHkA(2W zXG}#pBaHSAPWo~k`6<5IRl(sRNXMRYqq{$}>0v-tIV){A=Ew0;&p5N_S{%R{wy zamQb|$tmDWY_2iTEK&V3Ue?t*&k-eCPhX_*d2vJTyi6OZhGx|^Id_t<=q~Jc!|@#8 zJ`OMGsdAl;u@*%IHl&{K>dtDG30*S5ZN2NClF6YzNS#qn4@ae0zhh_Jd1Hu8cm7#N zB^!Lp5!1iR&w?o5(oO{E7e}S)Y>-1fJlkv_3;sT{LEd5rn~qLJV#IB=RPC5#s7EuA z)O3HZ(LbDjrL&S4}D)MCTB$2EUD5|hln%o{O7mnK4ymo{&YQ`@grF#dpzPRqFj{FV_vKh zBtqLW#?1lTTD_M*y|uirA1u|72>q#E|KlzRmX#1NK1-$@>>vcsXWAywL6I~nV-&G; zt)<=k9gVQ|7RYkc{_clT0e;}B{i_d9@e4=&P)hul2;d>o+qh6$ z(pe)l{N3=tHEerUc~A8nMYo`~KUc&pR3Sh_jh$ zjs+Iy1=#dUyz?|4eZ4Xn(MlPbDawBM8wu<%L6eQMZP%lm+XBr}tNIkfLn#SFW3e-T zgsem!>4RI#GMDOnmqNWvwHZWl>>*uOpIi0_;K!YY2SrIOnrC%NI4PYX)AYK!->y_V zDPapNQ03q{%(o}TwMy9V6EW&XCN%f@y_|Q>-gm)N^41yX_;Rx6vaHehKP_Yepyq6cE#Wseps16n`03 zQw`rBFoe>(qupARNf%KZ5D+V7ZA|F*`R-k9j( z9K9?Yqb;?aXtH5!Iw|01R}t(PQe$a`AD~|)&KB;y{lQQt=;-3Z5_&z7KG}IR^{Df9 zu>Z!q%+><&+OLx8tw2NU{9W^eY4F4ITov1n&RXpyJ=s;{aw-eCfXl*AMShpjFTKjt zd%TZ+Or&{)x=`&<2U^K>@qI^ZD}&ju%gwr_=9X~@wBELZOObud!wNsoxr9HQtVuEO z_%SiddN&dGz-vS^(sa+09W3d7rAdDs8?-~7>%VQevvSHL^ayPg!aG}3&}=x8j4-t` zo(gTT&$9^d4>_#vWfZ$kNM>gq_6x{zZ8dQHQ)`Ym>-qD%w>6fY?bprcqSd(MqL1HD zV`F`5VDx}w-OWWSzBk`}p3L20ATF) znYVC#wO+)X0xU6H9q@@asA#!--BcFQ6C4Y9+*o_?*AfzOF;h)6Ip9Bhd)fA^Qa-Z^ z?kwLP?*dFOCpxckhXhV#!!{&l$?Wk3`0$XTcsXg*k7zVv5#DRZJE zg6$S&se87}Hpsa){9wX^E+b?0F_k9Wkke&{?EtmaYrj`E@;--V{%q}H9u6sVv*ueS z!*ZVq-+4jL>oMC~k^0{EfAPPwc?zlX-SeW5lFh6OQ$VrnycLD_l>d7n=s)gwzNocn zZYbjqDlj8(q;75e^7YlCu-8VX9oflD*?d=9#4#c<*FJUVtWfPZY;;OcO`Usn_lM&9 zS$0H|c#_|C8hUK7`XCkF!mn8ePrJb`&7!{#&E$_m3?1Z$1f8b!zal@T`|51 zZ+A38Ju$v%OM{OHGtX}AYGoKd#UL`W6d0FC?_j?!UyN_c$$lF$CY<6gKmNa5GGQ;t z!_$VMS5UD3Uh|iDDuySpgQUcIo<`XJMF^S8=LvpF){!tubD2Sk4}E>YD$w&vwl84V z#G5(7AhVZAvh~b4ZGwr8wTT0p-*X1vpEu1Ce`#ghLD+cxFtPW3S-=jO@%+C8`(22n zcdajM| znH>#n{b}V%q0V%XwUXG0*3Y6uFOLz!CR(<2DFqa_@cvf5xD~yowtjQrS9w(mTFVGJ z?yZJ%xWWdKh0TcJ7n_=YWLUOgkHqoGt~J_BJ3*G+hTYX`&6R86=(>PpZX_L3 zWAWbsNxwQZKZEN=q@cflZ{FOXRQ~BSUra;UwOp{~!bh%ER({{>Sy03aBf@(eOWBL1 zyH$f??tCJhPAp=2o6~<9daipFD)X%}_iXH*GW1HBuDV6IakGN!Ygm-Ot7G7jf@>OJ z>tnTkofBT-7q~w^uhnZXU-%X})n+5SQ=G#ix@a%Bb>n7lymsT7%n8=ik^ks>FgyQ5 zhfZd*YQ|h&f7C4=Mdx1$bZpxv4JHxF&dCT_KYq7LI5-}mj20d3v_3QS0r()t>mq9egEKV8uqf zPkE|bWJ^wg$y6hBb3;ToBzTf$R$F_3j|K@;EzINe<(#+473-j(KTIDr1w|4d-B%Ow_J*ePlm`!uT%EM`i+AAKlJ@%@$Dk=<>A--X3c zNxM-Xvd1;sAgi$Bxh!-p#)v@BJjYe3q1F328^C+dwSBBO!5HL_Lo!*$1;au zBnxtpE1T1j&CYqYS;~%=9k;v?8X1eB5STNOclW&k}`2_!9EWYnyQ z#vJ8a#7@uO%2J-2_B7;0*mgSFvdQ20F<#=gzHiApM|o_Q%fe2U|G)dc8;IW(3^|p@ zCS6~- cnApr>AV=7og+nmD+5AMsLbTNwc}QSFV8DU0 zu0fMubMNudVbI^X$)>Ao32Voaqc5MZ!r*-b$v@u5WxW0z>FAQ&uKllHTMGvqv8)N| z1g}03+{2x@qt)BD4g)e~f(BmeuFvfpj32}PwU2l&6h48LvJ44^!RJ)|gy^cOep9{d zjv)q!nQ^gubin!Ut;BQ$R)Ngw6c2 z5$yc)c~@2@ZkpAbQI9$Lw%GmN2eA9tk^i!EZ!TT11s3#+T~3wC{ue@W>4?rzfHPY23N{8uuuL6f6CTC literal 13407 zcmeIZcT^P5vp9;wE-XvVIf(?xDoH>P5Rj}$mavEz??m$%TAYYyUUk?8&FNcp&_$(iV)6??P)2g7L<~MIf z6XZt|97fCG3np-eaaMkIoQ|HJj@8zlmOGq|CY+9zpTb#Z_2BgM<8o zB9_#%`<#|{*efC;9wMz9Dn@>3%jx8SeHs}>{Zl>cfj2Qf@5x*pyF6Sn_1*0o#qi^G zfL+y9ZFl!nUUSfW+}QYwaKBB!Hy zT@ppX!-628p#X?cB;Ywj7)10$1w+*E{}=y%oGril0E2XDp104C&(BkkDF52DRKu8& zbSrjBBO)UtJ3T4?XC|B6He-$rXjFYYR}~52C|-B1UI2 zAG&ymK>D8mZs4gZCc+OvqRfL!{7-^;^bs?V_nlAD4*;!T5Z+>QnL-2eScb;zA{WT3 z`~L)Po~UB7{Sc8nDwwkW9H9Ap7!7f*qbM@E`muV_g49F8qs8oZ#Ifb;8bnhYbcixM+9? zYu7QeFd#T>GJ~Ttv@B0P8WXNl!faqe<-_gLaQrs_+w#eItcxO`=EQ^3_|DVkG*SOzZ$JD{&{=b2FTDhBHv@C?{nfOiSy+#8;%}aulr#o73UwJPu ziLPyzeEOa~_yC?x-eOB|vcQhj^x-zU>9+-CZjm3b2z|eYthdmV&EyUV3RSi^{s7A_ z4P7+jo@ljk)ylY1toxa~g?~3ho;)hDp0X~$2YjSCLuq5IEOgnFEjT00Yf6yV{Xim& zX41Js+(?lif(`1HS0(JR?a2&6G?%sK-dcjbHPJ%V+h3k+_)!GB=k;@tz~004w2zvI zyX=A=HMLZk-4>iZx`N?2^QX{)SzM51Y11bvfh*SuqH2UTqtVK1L4m%K z1kqv9WufPEM&J`lk>IfhcqGQ<)glWNq@TeRq;zQH79#3t2Q!O*NDLlaNqnu3`tU+x zZAINMxQP*a`IXwS_)C9SMCE>6CyXB^CeCRwZ%s1h@;Q+XJbVdYAPxias9UXWO-a-nm(q$qLfmCcD&5m~#JDorN^Vxs zpwq@^1_FkK&gO%O7yF2NuPJQ%uQK=dAN?VBc;Q?#27wNkkgJkyt1MgRjAR_Fbd83H ziQtfZZ_7URI0yQHvpL4GfSQ~8vn?&up~6kDu{j5l`h_Y#cdysydY0Qw1Xub#hMWl5 zTfVcQUvjxM_;$^=(SC`P^;2%n&psfpq~E<2Aa^&eC*>_?7t7BVij=ESp%d}$6Vdg0 zT+aJ>BYHvz5(6_@ImIwHvNY;!#nHi9W{T|agx57tO{W&VcP`$ZA@GAwR11aa;yna^ zA}fzjB+_I}_CBb2TQ^8f_c2DE7xCjfVnWI4BBibSbxI1b1O=w{1{k+EDy1!RlXgl= zB7UeM2IeciU|H2=Lr+Xc$kW+^R3xO%s&&|`xqXX1Qo_2f{M}PnAy&h{lD8b~wB6;z zqj`XtPsJ}es!7t3Keq)w8GuQ$hQYEm=is`8vI2xsQk?;aVzPW}6gf2fG7sHJPL z5n!Mh;yD8j=W$b`WRx%;Rc)Do)cW);DL9+P=!nt=(_2CR7G@~on0dD0B+Er&P38V5>JgL;$tPHbn{n>0gP%i%fyWRrrLwdq}wXG9%% z*Pg;8Vf%7Ml55vTH4HXq1{CQon>4R0=)dX0dRLoH9VzbbG6?5qT_6eyb#rIRg5G8 zyz^ILFK%(77Rh<4=_sSN;a=jzXo`QxpLTM`S(qLOyOcPlfxb(~Vm@e+_t)CpnYCEd zz>Y1|?;owr93DQ<;gfv*b=aA5)+`blzMoYI=tn-9A5)jlyLQf?_(TIEX`lG+WIX~^ z8yNa%Lvi8BNzH&)R{mji%m}eR4U2BE2<{8Mf!tN2q}if?njLIlK(o5DL$X-Lzsg9Y6 z!on7~7RD^-{C@c{$U$JP0rC#n%b`)`?!Qxcz^smW4*e4k8oBI=aB-XF2Z}6jTPO&u z2>oIKNtlet2(X*sa)s7!ci--fa+0YUB}K?E4;BrZs0*KKb0}^)QIFowO!2wWZ=%`uQsjwYca^?T`8Cl6o>`MBo)R zwTm7VC!_Z;oTRL+sh9pOGr)oG^LUFGpQHR#=_HpVR}J8@$6g8XJ4)vYJbSDE6zWkS zP`F6ra0}skB54ZoY%Cyu7CIifSf;5R)2t2@n6kxV*^5d#@iR4VT!u&Y%v%%G;CiOGJ?ItG>^P#U~rrBgR#Yan<7Gi-#Mh_T`XJTh~PJx($a zSEbh+UWz@w^ooidxXT-w)e(Vnf$6+?jT-YTW)$uh4fnX_8+(2 zK=Fs>GQSo!yWoQy*p_tyCr@Ye{nA;eD-ko+yRPkdROg=U5CpaElQ!^dgCGon1(GZhOb^BMN2pk(d9 z7c!Ye0~LSdxmiH_J69c0K|Jhe){1xZ-kN$;Cb62BE>du}RBt~1ZkMmGCzGFD4H_6!^BTOLm||C}z2>>?9Tw zws##d;T6oFiU4=}+FmL32p2w1e9y76dLGFuX^fN0VUptsOXVxp*Uu2N>I%hw+;NY1 z0h{;(p0$i;Hg~w6LWy;Jv@uTb@Q&s)+`(hqYljQA-et;FP1 zH!~_4A`H5)gV0OKYJ|m8D9M+i!;44L1sT)uNOG|e_67T8(|9GrpU&$4yh;|tr0yFM zV*KtX`PSwh5h907j1r-Ai(7kP=8wl!E>MLXp@dmq4%b-cMWwk$o)!?<(d-cc%=oQ( zR*ZHulo}WRMUoGkals@-HCV)CUMD2aw3~_E?xSV~R@W~@;OJn%ZNxvpzOg1?DGK*D z_pM5+uWVX7A(G>)K)oYMW5U+Mubz17F<$JkFO4WFhzL{&B?Wu-s?`bCGL@6x%$YeM znDl2byf}WPBdv_E^B zp&+h)Fo>tQ=04kZ;lhv^ZMoyz9?IEg2$MtpaIHKjbs#=@L4Gg*v0^f$dTHsFMDNaz zdb(qcIRV3s5W-v(_vEB$-~}$Yp-AF~F^CmIzo3hv-9c6649CkC0vn4u?$~#J_=}?X z5!G{E(#cK$I~Sf55~{#Zev3juid*E+O<;Ow@TQ%a{f>U}WmA`=)LSTukv*nLJzMf0 zI5j)bt>uAG>h3h;dl`xI7r<0}1W&QZ+I4byudq>T_vDWzQRSkaioJw@J4o_-0GLWL zD)f{xAcy`zyI4$~OUodH5=AKk_@s+CTT0Pf=IW2dG0>yy1iIuxco)$&5=@wXVR&=_ z$U9~G_HkCkESubp6-eDtUH!{f9H++lECT9rRxmM#L~Dj_=QHkOtBV#h zz!bbNGbDNB74RG~M5z&OQ7FAyfPL4Nux;TtI9^M^JWkC3ws(z=> z#!JBAGBFU{e2Wn0snHH37)EcKu&wAwY9J=hJtly;RVGX46WieOhW~gIW_h>4A zOcMUhQ$7kUYMrkg({1wlmN63pUA|602Zr{vOsxAxYyvzIIErET_+|b_KTG(@yBGV4(%_Smo zPo5WO%k{jlMd)_9wsnPGuG@XpcDzL?OV?pOS`9Vm9d}B@heO05-Q9u~ z_7prp{yylKlojNE3>mZO%{O|lW^2$}G@UJz)PU*dq=oYqjKM9dgKY7AFS^OEiDEl5 zc~Jrkdj&+`%=0VSnJ&8+&IKdrd=5Po>vIUeieoGAbO?RQj_wW9Z(iTt+r>ZtN|PW- z;hwy3(40kbEnu(aq!q}7U4<(|V4h!QtWor?&>BHn93Y4Pi|P^{az9>Jps(DfMm`62 z_h28&F7ui3benJU7uf__meTnhuLWS#FH7Iw_&ID#<-NH1=)rA^E)^fOQM$?_(G2C4 z;O&}eVt+-nVo+)dd&AU>1;IKrjLC28O7Vj~B8#i$gmrGvS%tuse58UguElo;S@U2i z@89RcI?X@Tk_=Z7t3n7>B zEe-5+W{(@T1-6RTvzI#8LzNZp7I8V_-%K5Z7?SZ}1@{!5*N;0|5jHn-8hFXm9+gcP z%+$Fn-dBsrqB~h7LYh=JFP>;f0`cFenU4_{vm&U_s)(muD~D` z2yS1f11|0F7}HdK!OjEh%Jg}^P>h2p3f%Oc8$yQS$q<`g4j2(FQZr3%Ec(FLjx71~U zrKh>DQtQrKL%Xip*_8vJX18bk5aZraZ9kz4-ef=U-Pp_veW2Xe0q>Gj!t^F5InY7k z>V)x~HK%HivBR3{VDOe#XG(o~6vR5%Sszzcy?$$T`9scn2*RXbXUrs?9l@owy3Q(w zjmN)4?69vIU~0IfCeb5Zf3{ZSPk|_mB#;{1h2@ipyW-H{1!T6-$~mhF8Vv6Ha=SSx z19_$aC=RKQ2=@$^aX|KnvYc{!{(Ha+1GR%k26M*~P3seg@?)*dg6kN&C3OCsbJ4VW zqR-t=G&dZ&mhY4eCdS6t#6fT$+TTIrp>A)wite1#q20S&EW@tG%C49Kax|cp-tq2w3l8R-$9+tAa|Jh(h8`PeQB`?X0Wl*KBo`{vX0&ZHRh3Zr(d=N)+xvP+lL>q^G z8b5hqmNBgcTg?D+2T`ojWUE%(b7-9V7Gq`G@m|ahHEn*wsF4dcaW8>C8 zkE7G382NDV4W|8myMTw3visqdOAR`+7+R^%h%}9chOblE{kxCRT2oizHwP?c=%e!W zZr6Xf1t8tGceji({(YlnPfcG)-B}WId+GZQA8uf-z~gqC@t!EMkb-f1D`+C&1t8lY zjRfyfQkfG*be?#R^sWx919_DPu_9Ajnc9t&Pw0nE+!D$aL^kOvyW44UAa4s~;lbT2 zejR{K!N=52YUO%-(T-yLKKWRHZ*uz~hGw5lmTr+BY?a#Lh(JB(iBQATn0a+fY1r;Fgk_{W|e-W}CcW8{I(R+cZ?Rm9%! z${!wFD_4oo+JbXxg9>>NZEzA*{ZnHp)>!Yeo%zM~=WD#UL-l}xzOu{lKu!AjKKJBf z385s$2iXtEg4pVzhpDpM0S|3=gyR0*%%73exQFDyEiz;m3=O-Olor@*R}9!`WR;`F zmOaYQCn7vyyC)n0u|)$p)LxxUs@$YfOe422T=ED#upQ$s5U0^ z&P#w{sHe*Jb^46t4RCDyBBGpa-k1=HpDKSpCx+{GzCY8K7r*L3$(sJ*`t4o$=tbe1 z^MFg)yBW3Ni)-IW?lQ{38&M6%7kc1$4n@iOd0fY;=68ysqFuFqPdzol5vIe6K$Ay5`*A!RY^g}RHA%DE2Y7(H2q31aQEbZdFau!ANY|MJQN{l zGb*ZqVH`hgO~}n$RTcDE@7-Y>Qn%g>SQa>7s4N3y+|+8^@1-EDVMlhhp?YsnZF^Kp zMPE=#A;LAyUj=2kF=%ex`5=O9Y&fD!`Sp8p4(VG(&l=NyVf@hMS2rWkf4*qvyQ*BY zLxW-li)_1}#X$uTf}lZ$R1xSs+SQ>@MFoWV_7Tj~{uzSM+8UaJ+;WxFmws?p`o1Tk ze0nfl56o1&NYR0hRY0!X>C1#>-jNJgH@hOZhJd-NW%1s@?$-$#M2}mQ4s6%9W`ChZv;m!rmA6S|liP!bGwg_v7zvqxUHR7s`*wikU~Y zUX@+I9lqjvd%%PBzuRZg#TZtDT=6lAgLw4?DQpPEm##@IJ)|AH8%g%I>kr_2Flvi? z3pJ+qeJq|c@arXOM!A;0v)Mb$;(i zgYk(rrx>DX}Y z5QNW~K<|XubwUo*2$wmEfrqJs<*!$-$I%LQ&R5@d_RqBwC+cCA$>;3 zCj-tOG7lMMHCgm2ed&KvNS;3wnJj?C_L2!Qn0@BLisMEkn(ZkH48kGKFgcQuE>C*7 zQZ_%M#r7vqW9Fmb@Ev~GGaAzBIr!v2UR_3cY>IEFOKE0$v+VWlb}ki=>LB-Qiw|2@ z9vk2ORHjr43|ow#vcW_ZakzMd=w4Wx*G*4yA{#S3H)m5MotK1Mak0E>U*1cmN$y%- zY7%Rk@uBXViR(!3;q)_5tXs{I^A9j+RDjMOk4zTCI`xv(U6i}$T%$qh@%V>a{|-L( z=ZUE$%Wq?gu3$!N8G3-GGS|)RO_jdjmT6R+IUBgl@BG`FoP*Sx+|os@g52`7KnQ;R zQ@1niXJL8V*7yRlV#i46?{k^`1Y|9`_#JnP+1%!-nixV@5asyFgKtzwANE}SwNJY3 zqq8GGCQJcO0{d5)$|wGNwudU&oUzcKO9y#ne(eRWNi-mJ2a_@OL~{PXEG|Q)6bnuU zr!$f~cqejvau^*(9?T{0pX;c%oT*zUQt>DF|S>uM*k`+Iq?=k__%qQK71HusA1_m2E)_VBxq z;~DOjf~=Yai5v#A&DRx@t)j8^KhR4h;TBPk!lB$Gbfr5so3g-@R`Kf=@er4`f3UKw z8SIgFC~)p0$@9Crm4xqjE_76W(j$ENuaT5IT{xZH3HCSxS9mM`!*pnPrdkrM*jlLi{C* zTs?3(w6AHS6{y(1>wk2AoUpY@P9t|ot}rfTJbYk3-#JdJUJ>-(?ZQEvjMz=u6)}?4 z#mLP!WE?tyeox0!?hDfaH654Cv$N&mQ}(N)*-?s?=X_E54r@=piD23?qc^S;*N;V; zoJw!IWj*}KSZOLA4cr~(s<~a4=|O7->jK5ZAY}=?+{2wI>#4=ShZhCuhnlqHr%TDN zV+J2`MbB#!%F&h0lMfhcdlw>QBx=wybHphzv3I)@(Y0^f`2+sKFEAq%L)H5xFqEKz zFuY6bjn5wM8%DKxZbxHgzKYm(X2MT%b(-95^mlYc6$BTAz-=YBWV>V!x5-aM_N9;y zlMte?;$0G3SCp_^Rqv}4MFp^Znv+a%%0h~{RBY|BC4Ep8^5R-Oa=bTeyWAMF-2JM_ z-t=AwV9>42QT&b^v}*Quz0X26R((T_@E%<$)&IUPQeUa!Q~^++^K0$@WrMNe)4b32 zvq%u@OKoCbjx}YBfd-J}-nTCr5#CRa4qdCxW4VaSRu=|*O(5e?b`n19wTWj?`GDQm z^YQ(O-G6B+Z=b7>em7a~#uj9e{xTd=AO%LhJdQO`NGut>&@2@al(xf^s`+*)?Hni~ zHowA^UvZ5jRz1@VZi|Vyt65ktUKupMJOg(#1qm8iGKc?lcRhBQCm;MvjlUvZpAGd~C|2 z7gHzv57I4yvnjs{;kG}MgpP4?uk|hX#T_z#+GDoccUIHxvvGp}zp0}&a5w*P9om(y z@5ZtU3HvF%7`#&Wzc(bWZkuERbx34Ky240OUskHQ1n_xuFx2L(uCC7x9Gd?uhAd8!DFzCkP@)SPmJk_! zZtuwz5BzPpkUrfctqAeYpXNPm7!@(NkOuG9FQ5gnSFvUNWDI6gS{Lk)GAXU1I-#FW zerPkUCE$OIb_$c2I!f{2u7P5TqVNbHFqTi_($X6&5gWy*!OQ*lqCpsMee@;RW^4<`AcS^<5H#`(A3@gZKTD zDq)M<9#aKctbHm3eck(8ei1|EO#S^Vq}87o-NdhyTJo+nOdj9)YjIq_Qp>zo%8Siy zSkKgLeb-@$@aYQQ@ww(=fSCiuvo_u&36&K+Hpl}i4o2Ov<$L@7wHQ1ED_Cs1!vf%9 z0y#Wu^yCkqqhu=1NHT^fLw5}xc$5^EpgU5QWaJK`;V!^im<3+)sbQifU{OL^U|1CU z-8V~xHr7{r(5%~UbQA5j>R}jVBIW8`?64Msa`0{u52f{c9m|NcFkK9xoylg)yJZtK ze8SfIvzr@qo6w#gYBeI<_;!rMy3U-;jGy?hT6*F=c@tHg7zUUBlFqoHmVVki1Qhf+ zdJUX#S#ID*Rf1rY{_o8X=#}k?3wHa$mDc{gS{{a|*y(DkV+`!9Sna!D(X8$*d&`5A zGQ+pge^@i^6uqoqL6a0 ze7r?lT;AgvM9ATeR(Dvfqv-=>-@QP|N^|6Db+Z8_2ZHCi@26Yd?uJj{yeIL z69F&lbH2JK%Y7B&guj6=X@;B@e(rTd0`&+kb7bnTY029mH#G6(_&-r)HiR#D%#+eH);^fEr7(NA#Req`SHmi3Q$;?eT5K#+2KoZT|-i2|j#_sgp{ zzCz+n%CZB6B_T73LBe2?T7lQLNRi!kTQ#&!4w_Th-d{+g`?c_(Ke9zxMwghGMo0L;7{gs!Fz8pG`w_ou( z@|DSvu}y4*$De?{lTkC95&f{}+yDw8#T=H#r;J1T*fC{-2Xuj;)DlSw3Vr7>&K8|*0~`=R+u z?TZlFRrcbs>1Uw)Ja&0o`_MGp^ykUF_fqLk>G!ZbRYMs9YIAo?9#2!OkA{~|dz4>M zKya-W^fl!4U_DL`n=E=9PUH3JfE8Ienz%Sd?cQhpj=yxM;IJrjLs(NrME(<*rFGS8mA#vNtLIlSmY1?0 zHK!V9-7BEDPUG9au4UfUthux=eYPf#j7;rjEj@03c5-K8OgXNz8gVQbcVZefu_DdQ z16C|L-!(*Ox=3$(2bDjksSNs6cQ~q)FL-!t+n7HqRM0X~H8GS}*6L>$nySd4US|FhhXSda6%QL01j=r0elZU-qh#YUS=cGX zjpol=ggWnDt&#g$^NlNFl1bX_v%Ci3Z*!1o4WlqgN~ip&jo^bxvJ8ZeM*WKRWJPo| z->A#dB{>Nr^^wiJ!H!N#R>0u)a`MIg=yHi5cCE0eF@Md+%|&gGUPYUYE-qFvO)z#g zjb?I3aWA>wKwbn4bZ4u(7E~5KpHhu~FQKh`$bQjpgrTthI~IScEnT9JUwHf_%>x$z&MUdZARr^1QNP>#^NOoNrcPraJ#htw}WVqc5*Y9~BrZ ze8p`T-)3YH$k8N>SP55*+Aiw`Rm2$-8CxDVC1DRbo3+oj2qG1X;Cji}XBr+jVD6tlNd!!USoF<0i8&6ds|bBybxWY6Z2nCw z`g$EOPv~Opbsc+Y$ZR}hzH$#??e+HqTA|G<28!0VGO~57il+7J{v0j5B-pI>S~lYJa)fN_D4V=Ju|OT@vARi&lltSey3Sqdv3Qn_hY*%Ph!W0 zJ;M+`mX|+iYHnxiQ9t1p+K!jCY$6d}Gv66!47Howp|1?4+;DPE-9r}dk!Eo^Ff;jSr(Fb=n=V;||vZgiDzrG_Mv1nLLl z3oZQvt0aSJJuCyXWAkbsY2xvH&7;3f#$z=V=8E5=gKE|#<}6JlsOqwtji=l?KQE+b zc_vX<#8)5E8%}q?4%sbJ1@w*>@`D7TzLT%>-w8!}APnYq(7POWH zeUx)~xRaOaa9nNG+%tdp;pwQOe()(gKnGuM$~L)D_scw(AXqMrTm9zn*H8PJQ7ML(@5_x0y<|GTp%(t>7fN9Rd;NM@uHcd2 zkYwLX|MC8mkX-s4d%ax!gY;rn?h%F1oLw+|b^1DTqHa_9J-!cGe-Xm+$24YZ)96pu zT>cTqxy+gX30!X9Vr_n&rqUhX8s>fRlO})HPk_PN&GZe%=TTdE)BYT!DT+x~Idf~O zXO6X$YQ91lh7}UI7ZC;c7u2f!W`2u)u)cT}&tpYI~W zFh}CU{HQ_A@@f9Z(L3(qpPu}KQ8aO1IXm=OXnpF*)~6GH6f4}c8<6Ymx;|McT%T?P zsrT;32%8~LwG<GJ5W#h*Nq-O z%mg{$*WLE-gxAVh$m{O);88Swte?=eqjd%jeB*n6I-SeurTz@~ diff --git a/docs/README-unnamed-chunk-27-1.png b/docs/README-unnamed-chunk-27-1.png index 7b6f6bd6befccca02c0aab64c93f8626fe46479e..ba5262ee3509e7b2761b5e0042f1398297332500 100644 GIT binary patch literal 22406 zcmeEuXFwB8_iqTHH<8{!n)E8YC`gkgAiX3A(xggNN{BR(DoAhArH78R1P~CZg7lKm zL8?eE;l}6vzxTfP%Wa?UW_OdBbLQ;KIWx0o&-^ygPxUm&h#83i005bm=3_$u02hfp z0ED>MoCP$71^e~qsm?Ps>^A@)2LLz#{-yuj60skLMC`fAp=sWs={A9vH&G6dh!yb<6PECoh~1j7 zdsAZ5eByjl;{0vF)2B@ifF=jIrbO)S@Rx`s|8q1o%{R?s_uHo1+b;nDUxI>qlal%~ zGA633<`d=S6CLK8usc@FKV0+J?RNh5cA>p}0fE?_oV;yvxSdbDoo~9u68_4-o^QFH zRvlp*P2{O*?hOFYXZ<~Jw0y%p0s!m)t;dg^1!V8#ytG?r%4jM(7j2T(TbHvIaHD>` z*tzH+@Wp+HSx$bkPd2^E{^iE)ZDyLrVHX2CP9l0?I!b0d_8jd)$xfKq`)GKvUS%cu z-~kvPFr=D)af9Io;X{6Jfs~l5V^+E><2{ymb?5=ecpLN_x@Mdk1VAS|Lif<%4+8|5 zl97|XhduU0IDm7zyXT%p+1IkhDgXe69uFT&B_w^=_YeXQ#|Ppjin#+1pg~u=6?0YU{FU^R@+ zHCfAl_BL>&Bmi*1SRhKVVz^3utg+R9R5TJFVrvGf&cUWj2?$A%K(9M3ERXC6$&g2) z#O*b_)vc70O*}Yoh06M|GbuH<-&thtAY8bN;28`&Egf1x-eU9JE0gOH0N`IKP&+cx zSzY=GdUM>ywT%M^2puK`*pq_r+3hNU(14TAHhQnmm`L++%$R9mXEkkZr2eCavE~5U zC|_wquiNGV*C3vtJpDy*ds{m*T{Wx(F#js*ye-O3b{?0l(ksrQ~ zLi5vyzR0I)`Qrbx{iiE{L>6F1{|}E1F7&z>GdlVae7A%IfCli?0>04?W-;iE82w}G z9(OJP($H&$f)rrJFD!D7H7s>K@@l$wfX?h24$`oeG~bJjK9n05hVza7&mG3Uk|nW} z|1!M?FNodl&`$;13;uUM9rl~4o^ucZsM0U?!qsrXaOg*eJb1&A03F+nW-1j3phWb~ zJQ0T|z$kr|4Qpp=ti~8tH%2l2&8V+L6fm6h(BV9P=y_+Ll*z|oO>FD<%se<7Es5+D zC-w>Pgg5m7W>ixV%qdOHSu4xSsvq~l`*}UhyCHYT)1)P=qt{bq`qSJu<&YE6+05q>gK{lcAhfaj;KDcUaabqOf~5=?)H9U* z3f(DYE*0ypfouAHT>nL8<3~=CHkE2AWN!LACnwli|Iw}7h+h%s-4;RbmrP573NgY! z5d{+Qfg8TZhP)7-uc@`yCCf1%dx~F-=Qj!|5D>$e_{g+97A6cXnBQcC+B;E)ee$t~h8;!3)lJ!y60`rbO8D!&P zZ7wB|ovlHx9D^j<(Kz*Qdj-3wxHfyuZ1h?Vws>vr^g`BkD}gZtA7(U_(2d&rMVczX za4T-Bv#v`M@F8U#&L-n>ehe(FWGhBwcP-nMXzU#%4t(RtAu;5H*CS#oh0`3L_mbvW zpo=C8f2T1r%dL^pU%N>#NEZdJm*JiM_SdnwH(fiMceAsy7SpVVvcG*Ce>l0y8b{n7 zH`9`AH|8Gj6As;XKd5)EL$gj)C?|&3gGxw;(6E?0g_&Lp|r3px+59Y6V#0sYi zOhwadPD?6eYJH?FULgdLTRq@hd5CgFheYVa5zvq(i zW0bF^U5R(<52Zh>kht;({sytOATMNoqWaK-oxV}gQdI91>I~Gg2F8FSC(T(6vJ<-F zW#NPEz?~hnn3ak8`ufodD5t`-ViPjbkY?W}T*Q;(&Pf_Wb}-%0y1{9)n2ZyXi7|TX zP1gMqfiCfy?n}rSxRK=mBck@(IQ-BEPl_frUniwmpeI#^E&42ttWbh3T|h4d(bWd& zByLV7W#wF@S6R1?cH>)DUB8VYVY5KzBX!jHFXV%jjFQFQczq)Z>S`@z|FHq1Ek%3e z4a2QP!4Wrm#~UemX% zK6yV(V@~Gn9rNj=vj=c}G2&|Xi_1+i@;U%VWqq?0te_^)5^mP4?uNSm*6 zJ^UP2#u%NKPg}CN@NlVq%pLDDy}Slw~e&9KhIEk^i_n&8!cMRtfhM5@-O)JZq}tE)9tA7VJVmR;sMY5hFdBlmZ*6X5V7#ILs0`rRGs&9+I{AN zG417z_hUE|CXRuW3LoJMx8xb5Q+R)T<<1A)NWExh=ANkVD+sO9`e1fZ*8`>fG))-! zgWCi__^?@Hw=JmYV!nA_(%Q39_lqjQ&~?fm`V|wnRq0CKotLp*Kz-=|i$0;PYv=vT zedujOVeeJiHlE*v=}t|{n#WU_hxyY#w3vU1=o;|DI2#J{V+k^-F_FLCURW4M98;Q2 z>aFl8aHC9OfmfA_EyC{!JY;{4B9j%^Qg#`2xnqB%dCP=%g(25g;L49k%brL^#*f8k zU%M%>fN~Ro9SFhC9dl+c{~Y_5cdgd%h1#H37Ce4Wo}^#tJOW95Z2z%D64%XvSCTA0 z1Wtu>NhcLZa_K=D5!Ol~m-RoxsR^6Bu%kPHYNiVnO`I^?NM!-7ZtwqU%9;B{`Vk~; zMS3fc<00?kw?sO0)*r~zi-mgCJhLMo-z@T9bI+2F;b~*sMIu_~ zekwbodwr(pUve{7CSK2ACxtJ5*sQ>{wLP`0STITPe?ZXrV>LP%Lv2P2#5hU2F`|E%@Lc?K>tV8*#85&$GTju}gY{sF%o%q0%FOF`;1}CReg{`7Kji34X z_ZR(pLk1U?ytR}vD{7vVsb4P!{hT#V>l99{55Hhzz&Y%tLq#J$N_TXwTA-|MpVOpTMI!JWJ#H|iV{CfDVBM2{D&R87gvm9DILXE##VI6BKrGpcRqeOXEH@QxnVE=g5}dk z4HLBn64u)-*)9Gp2f|Wnt}feO7F%ay0ayQ)+m(j#ro?(= z!|{Vc+mUDW?Kf$_*!*au3RU7xi!wNt+(GlP4$S`uU@_5jB^cY_DleZkM?2eX=|R4n(xk9BlU6wy<&!#?9N{_z zXR*?H$BNFZh*VgX!X^;6RmanNmrvR{Ra+p5=@w}9SJVMcbV0M=!^6GrWhx6>9DBQf8pcEE_c&lRIRus`!i-`c+Q{m^EGoTdYPp22PuZU z%!?{r3O@|-Bua6;D&yIb6zLUAL-Bc!bUOk`3w185ms&SP|_ln4B7+qNI+{zqOyymf9{;eUk{_V4ytJR9(fs1>QW5rfwML&z5r)Jd=qDIoYR@o{GY0Xeg zJqh{7m`Jg7mXw{)D$K7|h~@KLOp-z!EP3Cl3frbhMt$)BALlm6>84+g-+4&DM#tmv;%p7%-*#fk` zcB${f?2cL#BPVc&*jOZvK|GQ*xfH{bl!egoH_~P`RzGs#tVd3r8!ji4n`KBWC%g%p z1pn-{b0q0NSepnnP7&(7f4opVwNeGQcz*dsWAP$64QWW#DoS0d#mM>BFv@RFN$7MS zody)r+OGH&&At9R;lw|03yb`sbNhkQqmUBi^NL z$i04T@$gPz)^MmoR35UZsns^cHVHM%%AMRdwhn^f(hogw-gnrXl6hxxYZ7e523^Ge zkx7YjA+cZG+DU&x*Y$$?5P^Acj40nXEv=2|e$)4?P{ILP9h2;Ris*F?KP&`I`5_-2 z=g33~=YUFNT`Y5}S+1fEB)YEVnFNa-o=fc5Y7y*ih~!+1`*&qhJ{rK2uy&u=4mzO; z5fR<2d4z%xCsaM*9m4$d(2MnoO>AvK7oBkDU=z=)-{jl1~mshrW3I%PKtE>n#^p?9#=}teAN4WCZCV%H&XNy zPo7U_v#N1===iE9j%(65KmdNo6aA1li{3I{Z$?T&&_SNV`O*F##$RDGbCv_cPM1W+ znV>00!H0A@^Ed3GfBY-NW}bSEd-QG-l!|ZGm33Ykly+T=h+W_&Biqy=>vqno{0em{ zi*PrG`?;8Hw^3;hcP_2ASeD!P2KHsa>N1S8_3Vt8&&kcB;L_s;h8?-g-`g9Eo!?O* zY6VRPYawNFv%e=OLa<|0)Y!}#s!HHWBzXbINAmAZB0YE&l@4`&nG;hv6MH?E!EMlFHCH9`T`w?NL^N?!jBN(Px+=X-{rz zL455Aiq02HED?gZvo58D^8rd%X|0G_BEgHcWG|DMN7jT3JOzvk_|_zVL@tNjynH*qGZpe>CkX$E8^8sKcE~xo3L5;)r#z0#C^0 z1%f!1!TBRC9$N`iOV$uJC~wL82D}R7EA4~%JGc2pEwwdWscOmB#V-k8;0T!(_Si;I zg+lZF3C8K$sR~{zM?}$d+I#wcvwOx(8v!IJ_a!y226aT9%0G%0_%=5KIUkceLUo<5Vae zHrvSO3eE2k2gS!HVDHsoB^6~)dgv{74qBpqlj@E_#iA+@-8#>d-_bIDo+;8InHqvz zt$B+E2rvjL(Cr)k3eAaMBGDn;SGBpuIs?b1#!)buQ+^Kc5oP+B|AJb&M3})7hDJ<$ zD@bXKD$;{a@M;ijJ8trwp`E5WuKDOp`!eDV<}Pg8$SokOxku}r4|1gVKB3iJ*zNCx zTN?baRtwT{IGFPF%JJj)Un^(((gv&p%feVk{&1TSWHPtKemE0ghv#=_`cm{|{I02r z#FG01=}*_Mw3=zgGhU5AR%$$K;lwR>EHr~>2Gy}D4}MV2a%$3Ylj_0o+BT49YRCQb z*S`732P`8%rt7#w!R?ZQYg;qe7iYmda;xQ<<2jpJ^P@1Q6_M@S7FwOihKcL-R!Ae( z$ebe%yOQ4=+eW;(50#^lJi)vII*3*96IQe9FZ}PX5R1=%2VmOzPA08Qk8`fZkHy&O z^^bo85@QW7MYMM<(mAQF$=xqT%;!vhHpzi>LA?CkZv$xuE2slczC`d>A#bmFmHQ4& z&BV>d%`iUL8!0w)RMOOBO=-;##K_T}fGo#I2PIjk6D~YEY|)0;G7eq<=;_Ekt-1lo09YTk zF5EJBsEld8J2MI{?ztcgnr=62pbFUV=j5}rKRhNSb$hh?HjYL?7cwfZ)t>B*d4(?l z6?yYG3=djL5_HE%Czq0M!}h1*ZyQL`Pj~R+Nu1jWJ3y7c3@7ZRV;ZNY3iJfXw3F&P zTL|30Zbf@xP8Zp{lP(2F3z~nT?%;0WZoi$xAH-Y=yUf-qJJxTzxmS>%=M`S(Y#q)d zz3T2D=v*8U@_YgVVc2#30MU0TueSVKZ9Vkr!Qo`P$hn@+l@r8?!~(X?$b!PT(zhck zL$`w45IJ2Pl3_~iUBOpvq_Uw(nwuvM zj`nmOct5>GU1Ds|3FbR~Oojh8yIdwh<Bh3h76r>lm4a>B&K5gGX>}m_5x5`ISfe7C-L2?)!2pK>EvlW!1IVUtT zXXg`6;It}=Lf6o@U0+d1 zpmvy(9Ai4|ll#mYA|h}_>qpe_ORlO)beDa%W%5eEKvp?Gn4XV!%Q$%IXQ{CnQ(yqLx(XA!MA82iaRG2YF{cgJUx3#CwaZX03 z89=An^Dl5`ny-w>UO!e@f~-pBwp;-l}wUdF2xjoVR7@FYfCiTcoXDBv<5L0P{CYh=$_W$)eFjOU6GCn`chn`?; z-^(Y;^n$63L*4mEQbdsAmb`49Agy($yklkjf|sjjRp(sbOl16`=69Rz1VPCsfw5-PrfQ4fjzaGf)4 zWAXL>+}&squI497Vq>{8qates@}@&FLc@qEnN%mp%`SZ&v*SI1u1fVB!ZaOJSf-85 zJ^;S_z=a7(#(g^YU8IqxvHf|@uaq{R^dMA2-7s+hocR$+KzjesEUc733h)n~bYx&~ z;w!qRpVQ@s=`Ezg9w>S9+e23m`p8+{t#f-0pHp$wkrTRx&3?rndy7~WsBrfGE+)Ab zs(LJx$s-+x2!0M^6ZRozhu=}+go%XXG<-*W10MWjIc(7E;b?hC#KfVcHx}0W?V%L# zkN$+$o1&5QLvzA~*{>9zQZSM7qZBp7wDfZ5J7O0IUi{Hz56HcBb_!U3r%RV~kO{E# z$h76E^6>25cf0cEdqU~aW(=W7t{@&@^Ww5-sLAjIekFeeGG7j2jBY$1T<4kcq7(3x z4##L4DBKvG93<38@ox6ZhV!h>=Htfih)8>v?1whv9K;jCrtf<;;EI@{Bj?EipY9TD z(&fC47*$1Mf_clXC&8fpz0e1`60YyU31tbV=U!2szi$^>@4KztBVU%FlKFs`6Z7%s zjZq*Fu>YxaWN}S%AVlEm_L@-PDBn+M*_$$e7Yxv-DoLuoEjV6$)zDXa4{Gi($M#k! zf*IJnEO942*3p@Oxj8j59lNmAO(^F%6Sg4UqoGNd76GrJ{PqZa^RFmWY=A>YR zBl%K5uuE?Sr{4k~M)e!yCoKcfVepRsS6FHPfcTU!FB-XC{hF5hkMtWBXHYx}hizq1 zIR`g0YE0y4;32x4L|AVGb9`DOBBd5fj&%Y3bC6n>s4#cZMU?9MIZ2Z>kK_ytGz<~_ z@&d?qS~DXaKYH&xD+(E91FAINtcpzwgzgm{qx{2Uh{XG`%;dDUZ`7gkZd{L{EMi zwhAl4_g6d;O+{Haez(TV9ybyN8ClL(=U0*{#r^HSG<3+?uqSp~+CIqDXxXK#{-v;B z>2W?wP_TD{HeVT%NZ|9{X>-~~ArPzn0!fqkaihD#=cwTvWPe-6^4LkF^UWk7DO4kQ zFgAvcyzK+K=+4`TT-(7AJYG38J98nj2!O|ly>fO_Y60VdS$3#o zT`XW!;AI~>r$-PUWGEynLU_ayw!_qI>kj&}EJ4|_O=dLu^y4eK_Aq2uk+*<)m6#Ku)mNpa`3Pmu@jZh8;cnUo9)3!TcFk83!=NUEH>Vx>ItD1O_(2N+a8 z7~SblV58~n=vUP5RG8XSX?_TSKU}DRt%i^4qRroSgFp3LyBDFLly(<1WO*6LKaV4j zw%Ik@A z8w4}=I6f=-^9>gk<`oY7bCxtj(KwHIeZ97GHcUdccp$rurMBQg|5Cv;ykN}D>_D0a zN%&k}~vJG)=wK=N&ZaF%g7#^68 z+K4098-;=GK1AYS@K?`4?rbjpi%K!8b2Ftj(zv@6qv_4=31b#^AMo*kQ7iO#)q#uq z(fiFzJr52sF9h}vkNsDtqYecG$B{_XQ|1cfqLRvLT_vWc&9S4&y`p8xtOy2c&BsUm^_QPvf7W=XDj&f{duoMwO(MO;1~8 zWJ{G1^D>JX*ebk@cohz0bI;sC30$dEzPaf;jJPaGnrI`%HWQMM?UPYAC>IK7z_V<# zT#^pxFi8-&u5O4W` z{0wf=UxDCZhM%4nq3^-D<1riVW?!g<&0srv2?@Yqln7#fb8PkJu zU-{*U=2(vf)3mLQ&wZfZ#!K#J@|_+?;e$FvruRo{e|28aqsz^)+1~jR}U9?l51eY5ga! z+Zha+Vqx+9e+XVBJcISrb|2YT{Hr!H-N9(jN_cx3?dfOnwm$JDYutpE|5J8#D83IgX1{CL#j7fM8TgE2ViKM6_I6pn5`kwplgs^@Li+ig=UdIqI=~4Z~t%ZgV-dz4( z)ll>D2;e_XGkb@J>^3`A-NQmahxFIZgO4;G;Q|zpD&zq1K3()pf(Ifx`wk#fSqg@_ zS-#r;A;$~VOZt>ck!LEFR}{p<6HM$^u*HT_SXtpBzEY(*S(9jLP<&y4ayOHILVenb zh&~c^o}s9bQu8`!M}XpJC+{5^2mkTE6~A#P%4!vgPLJEcN+$-INOo5u+u7Y89pP>iCRm6oOUHAiBWR zb{`t}d9`75RG<2^{4@CSdiizQVv$<#@CZ{`_mbi&kc-$d+WWSx2P3lR%=!G;JQlb| z;afKCcO%Rt#8U|y&l@$)+=xUhW@8ciYfdHM2>S|Ao@YZQr?U6$uT_6gX&e%jx%O|an;1#|$O=D;AD~VVk zDh-c~ORA{g=4Ih_s>fBppkDm6vTpJOUwK`+rhTWglaAKF&8QiJs=ts>-=#ATDLxMn zxHib9Sl(^nkS=PwK)U`~t4GQ|DdZ$Kfkyt9Z}9aaG*)kQiWf&W%d0z6e-ocJMF!l8 zu$90fN&TZg>)V&YRGDP>VFFpLW#8>>4?=EQyV)t_qf=u^MqXj0JXdu5?Y5p7Wa0R{fidh^*ZTnTq;982 z*oOHI6_`ID8xf`3J*<}_e8xX_PPgib7HmRzs9M^#;-GV8<;dvL4=xTFUm34;bGJLS zc%$c(?~0vpO(Bh27yObLqo`lFhQQ^y*Ei5>(yv@LTH<`p+V79F-RkFVmuT_6|K3FH zQ;{?=?b*F>8na;d_(7*darDKi^uYy+i~YioSlFp`i!JxV$zGYnq8~wY!FwQY&GJ-^ zq-|8ZR0xA5Ug|9`&On;stJ);B8oJP%49v%SyEM??)Rxa4cSVZrzr-OvfAG}J9N$*Z z=G&x^xp9cZ6H& znC?21Uw7uR!3%>q{B0~9nA(Ju>%r^#YTh+hr}agB`CKVN@(iZWsbT#Y8O1j4vBkTZ z#r5M$T5~QB$-GYXkA6ZfAr~IcayVP-LT)xM*;T8+>gaO~>vE)of?<@+r=oWwigLG- z<}>A{F05sLBu~PE7_orEh($yOceC;X+s=4a-&!CpVE4ZRoBvLCFeJb})&{*liQHxB z5|P4YMdZ`9B!(gZf}~Z*Jmu5?APmQhc5nfUqV)v+dy@QH>YvH;FOc;g{OtdPnEtom z{}oHXp4D#(DfvB{2BIk(kd6GEfXRpGRW#3fjKw24_CqfK%#;|UM9zNr1cvO#p={~E z?+i2vI$1V0$PT)|j9p*NUY*)4_clRhvBe&8A0Es}$13kGNn$>UwLlxM!_Fj1Xk~C= z3Aic9-kSsRt2DK%1LGe3W^YxA;DJc)7|q1OxbNSWy|HL^I6qPD?CRWEceIyu@n7l zs*dSPpc)J6O8kl2H;+zTKjD_e9n!(RsK^CmD7 z)+s}U6^_2|*%5fY?H|>nKU$l}N%1#}|F9EG68u*8)*OIsjwy^H%KyZ@C@}9vTw5F_ z@_(Z1|FZM{${hcn<^Q24JGjF(<0&3a4GBIqKtce0$qRWZT)sen8RgbrO!@!ynE3a2 zoXn%BVWr6ng0@@@`x|ttLI6CF-;TGfOLdkc*Olk|>H$~~IYWyKEjIa2d^~}Fa9`2ajQHmFCODGqR5@f7qJVuvgLwJ+ zV%cCq|BSqC(f2PiN4oO#qUP0F`wasS_YT`&F!~BCZ*#Fk-FxE!#cUhDSaUvW*5LP) z^#>xwY$;pr?hZ@~_kZDMl`4k@EHRM6aNZ@%=K1GI_dIXwxk|fn=(!oZToS{+`haHw zm5a&?d3jCyqfzQypac05+T~9Kd!)jAJ&Oo*_f7vbw|}EN#o7|y#m?c5*}c7N53Y+p zB@r1=8Eo53>B4H_yyf!GJ|Y&#R=w?a+veCi0L7Z4V0BI3pXHh@((*}TJs3W!tRUe( zqk7&*WTgU0yxXQ6FY!Z~$;I`-dnFT?zbJiC&x(G%c;S7kDes`5KjHLk(A9=pwtp0< z62G#PCU(l<`;TGZM9dRQJ(r;B>j)$}hdT5WdewXE_3m(oDu5*Hk=)^{`cDrbiL^P& z!y(sLPmoVo{y4a&MjL$>7}kIVgwh%|CYzBem`thf9jwy=!|WfT6Q2RWR#u-#@CU{& z|J3FpF3%;nyJM$*fjU3KknA++Ux_ub`gpEzJ$uW{as3^%^azqj9hH{li(b?;c?0vJ zBsE+X7G2VFDrS0A_%A8{5fe?1Xd3fJ-l6AG#bPkk-@KW`f^M)nBig!oeRKZ1u%P9v zWm7BGjZA=sp010Xa(^1>)s3xg42e${J0BKZfw;Z*%c0wsXu%gLPuTwS30ZBJx}p;8 zD-bSFggFQKn_r-ds2lUibb%u+M@Y9~Ne6TF22k%Z9NT$K5568cX=(zr&_l8tmu(`KdY-*YW-UprdPd)Y$82 z9hH|8ba6WTXJt)b)l?%AoQHg251W`P)Oy(c7!+eqDFcaI@!MC}FyGpU;4>YzGzxhq z_SPm1N%t7-MAzk<8KHXpSc{PE8^32|wbGdFq}e|1vgcyliZt0Yi@Pz;>_8yN@ADCkm013xhyM6?A0zch-`vr z_w(UrNZDX%*BiE$Uw$6%S1_}iYVv6`GVN)|0n4gvzzwe}h{cOEr!)9IXAJ9OGTHFV zfQCgsCYxnX4aVr+ysVAOGyW+RK?<0oQ^WViPRd%`@C8s$zXAUK>>d}zy~v^uB`j5U z0|a=3!vox*IiEvZu|v%yA?xp>a1gDX-MAUdQQV9z!wT5nMp$zB^;Cmc;k-ywr6Dos zlU7s(-V*~M;?naAi2dp0#lgv$GILh!>yoT$&1y%0$&84i0UmvZ_=!f1(^ zdE4e1{U@c;saMD|M0{Hzv>oY9Uzw2e6yi%oZA1?bZ!Pq?-@W3AKkka)I*jn#C#THWE|ipcGO zGCzl$BkE!D5Iw7DQwp_T2P>;C^6dw)t3UgH95AxEJ)9s1HTNsO(6#OHr%$Yx#&Fn* zgk_$pHD%D9)Jh_v#qgkQ-_p|P(ugB?`&Yu=ciem&`p}=kM;c%0+49N&a#HGjzTNEA z9Y{Qs(z4#ROfmwe*B9`FXS&vd=e-BynF}tq4&uqyJ6VpazI;p;<%2%|sbrspa>h?9 zKo)cv(2yGoKKW;MWF58@f%O8VgC6J&q+Dj2kTBQrWZiFb6By+-4tR4Vwy@;x-GAM& zdrV_RXo;4zXVc6uIK!2qspBrw`$;mbLH>;2WdS>_oLnp<{A762$``156ZaV+-D>o% zuHaG|EsS%c21gd5!w^50+3_vECtVC~5AreBA)UW*j8KOK!d{2<>o{JFvB_xWZ+V_< zE<%MKqdV7pPo&G=F5cI!5^2Dd^68o|x#tO#h+3zP9{#aCGe?oR>U&CjIsSSvblv(1 z#)ZsJ35`dYPCq2?N;yvOW`;eYpl2X&U+!FaV}U|pX5NZ=8~?|u%S@aFWcWp?F=X&F zilb@G1IoeArTK%^n!Nhj5c;t8Y~6U{61}aa{Pkm)rgg~EH!>x-?gsq%Y4-y_T+qxhxo7nzIeH&xtczNfCr)J=wzDsyGD z{u*Um5D|XMXhn>%`9^$A6Xt!O$Aw9u zq?vj4(WV`h<+$HhQ-;k0&-o9}oK766-Zn>Vi!2vq6bUO@hb=P|D}HWz*7})0I0DR1 zmYZmdIWm=M8P2y$tds9b6~@>1HEZxUfn`N7_jb^GJlN?8BpPb8%y<62e!rn`Kv%mt zwN(9Xz}DLcd}(pNiSmKN1Q|gtO_#K$Pjf1KQTwL(HJ&Oj{@hKD`NCD;JG)NG`f8K+ zu)2+GdLwKpj-(Vj5Hs-Noe#S7A-2R3^p$)XbMZh)AwK~FTet~&sSHEHshU=ayHv+B z#LTmoAbCbIB|dC051b$#x)AWQYVS52WXmz;qgtSFTN1!YDKlZIt6_WP(Bf*qq%dMz z+)+>5emgH>5OzhV)Bkc$`7*FhxDlOa5+W43CbFn(R#zdsasUdOHwo@yNlKptXBY)* zJ=p^aaMwUzOr(Xd_Mv$msPqMdMa_El3#T)YN8^K!&VKguRT}xb!&X;?MW=a0T01qt znU*K*?_0gTJZO3Z@nTFqsX2bv;r^IE>cOYXF)IQc7TSrzmLIF|U&TWFYENIO+>saH zxXS4=Vw*H*GKiq!c4QgOH0n4>L%x7&A?vvA3%--KUOa_kiqWud|m) z^M@AU7mlLo)Y(@^ncGVa4*RhSd0SAqit?UE>1G z`@Lkp?&pJSDHAzcB91xUBay1(QN{0?OLcPK26lA5j?^u$@n6VCH@-{DBCGl=lvmwB zf9xx2wp`B|A_S;`n?ROOwbq^k?i`J>E13g{%yK*fah!?lS5+VKQr(#yYEK@`hq*N$g+GOoi*IXx&-%JV8Ktr2 zlkVoI@qvuTQTIx>ZbKxJmYpexWA6?T_kk;rO=;?n>Fzp`jMAYYh^N+US*j7qB$x_; z>a7RwAtW9_Y;_c@Qiy;bxjXFL`%<|&a*MfqkX8;wQ!T~PmXXOZ&SjHL`V6rP6L>=zf6pZx3eF;4;Cb#try~d9=oK{LhYo18es18p5Kb=!yN;BDj%#JYs-!;|GjOvU z>d9F~zbPA$FSk3D{&3#GFt7dS`RW`w__YX;k{UX$XP&4q^6XB!hqAo8?Oi@umz1Yu zL6hf13t-J!Tq0l$RGDyS?q14RY2u|x)e{NF!&Ih1AGkxcx$F`+Zp(3bZS%~Di&owi zm7Q9IYmy`IxYlfBsGs?lhD_oNSq#3sW@S=Zy}twOW}_ z?1SPkE4C>rs%DtaXn%>v9W2-T^sHQ#y$dz+PHZ?cugj3JGQbnhZR7jry|k9LpFO+_ zPfcFdpy8slhl9@EMa%BZ_Xx{mQdP8jKTtNn6_Z+}=mq5~KbVzOyB<_x{c(jQqU&Xv zoZ67f9LoUgt%{h&m)k0nj*stGuQ_DdCdI*+jVkq+`Sd+tVvXf8+UBhJNl{=xBl5@A`GIbAzd1k+sHxUH#u;3iGLZL7>h zw;M-4Mb?wPA$Mw&hP6XcGFxXY1`?64N7Z;0}$6s}H@%K{l{Whr5^(0HEdg`@aApYl)i` z9@4^8kU3L4ZNx!&>{W2RM{&fJV%u&(t{C zv& zSU*z>8xU*|&PlFi-NmjKpY$FqdJLUKRKJhIg8i76iVxPy7@#Za(6_#@{0N#d0drZbwAVzl}xU znb(G;paOoLj!`AY%OgEewI%1_av(`s{WV_Y(NyATrs;2%G#A#mUhE6Wm$5(Gb;OtG z-Ybz*Rs6_#znm;&-u?rs@+&sbv&>68oUqa07oJU5ovw_xMeOTCFX2Nked^ z<>OhAQx|0ehJ&Pq9#7YO`tLNKn_-CUl3@c7_gUJvH|c5iu+ zwOjlnq(7C^m#)1qxpN^?yA0P)t2~jumLxSsM`7<(^iNi+p-k-DpuNnKl&+G}{U$1o zaD3d{<*b_t_D{scPp(}2{O7rK2=`R-m_Y5vZ&i=8?2{*+;lw~>nXO5;`zc9LsA-*d+$Ejl)zsm;;)A!drPly0cS(6MzFcCa zwf|PjnTJF5{(U^tjBO_S&X8=`m7Q!!C`;LuF?@xxj$|xZz9UQ#NDU?)Zb2@D zPB>p8WqbXFU+(aXCFaQ_4-7$SEHZ4G{q3DbE7WWz!XW|kPjxX-p&Eqvd>yIz8~CDN zK3=carrZ~~obLOXJ*Nuelt!H0x{N-_Ouswvl>cE!OWun}`$i+TW!AKkrB>fdlkwBW z0z0_3`M0j$Kbg16i@P-O9#WNycuVYITr`099FA&Smjl$=YI{Ub2__suYe&s;q!c18^i!j zbQH;yKa9}(;tv!T+fB-1QV1&w>3+tgnT`!yj#9K&$<{$b$72^lJqM}k8xU;T2L@#Y)RslJ2 zf6c`6S{$1efEmr*ba;)MeleAZiX(QaOD}PI>SQ4B=@)fkzr~~D)V&5a2y9szG4Ff=l(du1t_Z${BuoIgRX9667p2@XRl zQv7&E9!Z9r1(FZXW*lXFrB*^adn*PqQY^J!iQ-iCC5ItU^yz(g=zu`a{b|97Qqk)* zpPYHN&`wgTOTZwmDcsH(-MhxAJuyY9T6)^|OFp{(hv3DOdhN=7^R#-m&G3Ze0%Nef z-hm-kmOfd@unzbB@Eon{;F|4arSL)K%7tgxR5%Yh(@AD^AJi^YODrz`lK-^i%*F_t zQ_iV@Gb|ki=8DG21?;m{X+6j)C>!wG{F#V|7*bq}IZpdg+@P7!1qI;Q9SNCxa7m30 zC`QJkZ<{DsDQ4WxNsDcu+X=W-YcHr6&3U6cL3?Ax==;uzpgK!es-sTG&b+18P70sX z4Bp|doW|cQUko04&>br*h$AXyAKv*}t}T;~2oMvAUrho&;PK6b*p_K~8QtV~%HNb( z&A=8iZvB(`sV{g>-_Pcmh)wqsAYUFnb@bKHOI0HM&1031oMz9bQ1Cv3$%Ur9?v&%d zYMJ^=To^@+LJ@cr)K&_0SX5&OVn?X>?qn7po;al%TvF+$$_%CLb*yPdgU+cB-Tgae_NA55uQK>!1TtrJbZ&;F?>eu zJ7yj=b8()`2bo1-yQRM-#gP71J;Ctv*D1aalMCa8t=tcIOW)8KRqgHy0`o>d>2EFL7i}I(xGwswxpu5Tl9B~-9kUO1c5IX zjB5a|#AY8XUysJKWsR9nFP&J%^+D=9=Qnn**0hKU<)o;Hd!YS%3$b%gG?Nfecx++M zTp5-OlCLM1@ZL{_Gl#K^Z6+WDa*4~CAPpUd^5;!O&EnI{pqdXtwJ&r}`wfn4)eY8o z6A$LRpyO5oiD}K)^PPGdBHxTwVwv6$7z^_+z=iF6!q!7OQE(LmDcooizP;rZEK;?y zaKazk3|D^BH5uPo%mri_`9#U2M=oer9=|oLTiuBw==N)7J%ZzNh}b^=l-zQG%ukb# zg3xbQ%$WRAr~FhLV^?Ch6`sf`DAS~k7ku|5LN3N3pAx6XPKFJ|jj zYeC3)F*mU$j0RflaTJP55EPB*eNcBM0&t_q^7}SfB1hPJ6A}C9 zk7eom@f1S~j50XpK?}yTk7+sEmvWiG7`>gZtQ{pm>q;&Ro!Pokd6m5NRSuvBlFrp} zM;N{2+D|^E(%g9^t9U%#B|#nhuC9{m^?0xM4qYZpKqH}~cW+(q^Dl5p*ED#d+VwHF zR-{xqNg!K+ql6LL1V-~X?eN6jGHS5pF^Cb?U|o(iMH8cZ#PQa&mn@AQAk}*_tdU0i zG$hwOQffi#Pr;48wgTOvz)D4(rMS<{xBs%ZEsnoSlVVZuk4chM$=Ud+9P|ByI0`PtjOm;-2UvrV$2|Ag@i{y-)uQ$ScUDNtz2@0NpxF2l!Hv&93kH5+ zwSv%gWZ9hZllWj-E6bO*ca^OsJcQ-gvgUg&oEQ6yXVZZzlBm0-kpOfpb?vdlfGK_{ zm0`$sF4aU56K|g|&czm>uaibBligFa3d*_jJ+&}i#x+%c^y6253^6?|q=l4S=tN%! z)o?SaYguyNw=P1Dhqesrl2=e&4s{SvFibx5SxY}yx`i=D`<~-8?$!B5mDxA}H^QLdtDb@!C+9fts~-b&vd>HY(@)eZ=H1%kE_~+zGl3)0G7`2o`#M=xcRixN%dH#pkda-S8vEn$mbuJ203;&XWCN}rzd9eWRO*n z#xY$dRo;sUt7Z|*9tfTb9sK+VX!0i+*xQlmADHre$y!tVveXp4Lb+-pQ~2Yl0xIa5 zX@HvdWYmM;MnSf20Ln2bg?XThjkExAL1SDdy?8>P#`c!w0(=UmM$hL>|*E+Yxt=etFYU{Z3M6(103Rvo^6O3~sRr(@)x9ubr z21#dFp8Fh+5K~xLONB98+|2bi?PD`uNHsV-yN^Fy6UK6Cf*CQ|)j~BNTsPi|0ftcL z)I=#*({ADP2<*6yA2SRTd_(<@D^yvX`0}Kmf@UnPLGkjzo>?#-?e9&Vo!h|fwfT_c z8($fQFKb&_A}j5h50`rlKK#|drphm{C|f;8@cB~DwYapj7{R(bY~78MX#}*q%to=i zP)2WM)GkxOy^|u2@sM@z!)h_N;jY?WS2tL}ZLx_8+EuL3S@t6>uOfpv@{Xh3NzIQ? z7h#u>c6m;LzfHLPm=W9c@X%iQdw|Tj#xmC%V)Qh8Oi$j5E(|D(xmV~zJ#4*`JJc^h zd*Vk*q)tq>-ahD}M4-mTZy%o5JG*H}>IuVug?);~S_3qaM*GRK6M5K1u~Yz5E!l;G z_ka9ST+lMK1S}w263IXSucuKoTGD}T)E`mSg6@)h7xuKb&Hc#d%YyC{%25u`62%`E zdb?kFcK}{sHk>Vh2biJUzo8SKs7%`!1~ag$GV6bHo&$pZ0W|(WxPOLpJC>5w!5H6% zIv;i{C8vxLvW7Yzc`W5w2Pht|A{i`ovo_W+VOUhl$A$S^u*;3GO^i4-d*&3Wf*pn# z7U36kjdEjd{7Lvs{dL@!tln;t;3y^l!f7j=ZQLbtgmni!p8Cp^K=e&Xg5mU4z^?yk z5$#d{YzP0p%HjWT_s;1=F@5F4>5foyj}>I}del1Yh<-aI1>HC1-JY#MrnUPM#sPo7 z5b|QtjQccE%+()gN_O_>?M^|Nl2psy|x zc;?(ENOX6P2nDeFn>jBDbz!;tFVgLS#9w0&mdLMpRps29Bs=kKT#^y_#@e1$bK=ki z3tocH#cr0F73UwDrHWL!h{@{G5o{$uOHh?|CJUeaMw1H|n`4fYc*c_>*Vmy9 z&C!q6tazf2l1GgahFeKZ&N8sbZipElt(>ew&k6#@A0hdaUoKJElGEcgy%X&5zzLy6 zR-y#qOESrK8+K;*T*+}5oc>OJga)mn4x5f?CAB|n5W(Qgl(-S#;X2!Zgdop#nxo~? z5N6V!aCucSsDcCe2IE;zv`;mi-?N2O>5fK4wXaORQ;0#%t?vM`hd;Z3hUs}doV31B3(B->U3)$-+?iG_B#xT|?sK=Nfvxl(*r^Um zej~5K1Xpu{iwMidc2!76lv<}q=o4WqkJJYdj5i_KR<+l^xMAKTXyV`yCK#9aTaT!TpGdL%)I6COvxXhf^$_vRWEV&+R0}<63rTr<) z=~9QuWr1c_(3bbEZ8p(yKf`$))NJ0Ao;IgNPrI1nJMhXX)l+s(Bu9|#t;LhpNoa$O z4MbYgKATns-Iw0+O34!(&}ij32A2_J z&5k;54ebDpE7B7LXZy85WcD7h^zn?68JWZ$?6)BEBKyJ`b~EaLtJzKvSK70156s1{ z#jS-Pvm;Whd$^gR$1^s(-^?hPlbuD&jFm4Bd}?0KZ$I-YU2diuH2$QNA~rV&RHhP@ z-ff{xFJsvV*&rlw2^=Fc9oba=TZ8z&AhmyJ1^y!c zuCVMkY+25*H)y69+=Q(l6%WY=56d7uf2cH6(9vh_*-yV~uNFKwY_e9djXgN@Ps9dG z^AnC@WDxTsC+S0xHV2WT`*)P0#S{dA3wQ+%*d3uymdi&wg%S_?c32T8zJF0}w+?I% z10_(K`GYIhhY74bAmHfJ-i3?#T9`$s$3@kTj8G%Ol>*nLPN6i@A*f2&9Sm z@l4?n=MitlO+&?Fg1wFoxJv#J&A_xfZ2|CVm>0y&m8sH6IGcf9`}ve}N1G~$d(ny; zEE)%Y5H}ZyzR1MahQ}*WTU`nr;HLlAI{KeD*3ZM^d00rd=OC@_7_gX|SY4<(?-u)C DNFE1n literal 22212 zcmeFZbyyo+@Gl+$!QFz>mKG>Z1(&zb0)+xC#a)UOcP9<*?pmnOqQRX6C|+Dkkl^kP zA^g(!EBF4M`_K1z?!R|)cEj$>nanw7cJ|Di&l34oO_7Y4i5LI?kiAxVsR01sBCtDv z5Epx84uxUGejUDjqxlN^4FE_009F7A34jFlm;kV=`@305NF+#DSxH!7k986W z*n+hBm)lALfTdvxRuTz+>3_Eb?8hnrd#tmnLt53{#dC2bNB|PBBK~E<68;jgTOGF4 zCDb7kkaY>jyTZ3`>#P8ERuXjy*kbjUh$a8K*VQ5Gkl1opcX!w5?cM0_-;2YCJRg{F7TUap_TTwP{`_>rZ()9$QClXla)RS&}nL5o0+3;8vK>zXkD`@sp1QbwdMr z8$17fz$Z)wo{X_&_JPGCJirM17X;vw4=xl(iGh&xpN@Y_0o2{9sLr1H10dI5ayEh; zgMMEv#6&X&_)Q*-nADH6GJ2-%0T2@X4K&f3HFm1L>B&z2Gp=L{1c;f_Ult~vS24+$ zI31Wh#RZ&na3AhfkIk9m85R58kXqOPtioQQ#{QV&-JTLcaWd|c1E)D`NDFW>*cG4@ zYrnLf26az81gwX9&|eRU@bU*%cg30ltiso14tI|6okwxfRm7daM4uYXj3PZ;kB!7Z zw>Leq3a=>up9yAu1O8Kgp2NWsf;HEEq^v0+8DbYk=UShfm&SjsV|u=-(lh3hDkUnG8!w5Ta!G6w;0) z!kR!!F6IB=&&=z+5M3o8f{-&%(k!jr6cG0G2uXBb2w;(MxfgZ^dZNT7-9QpLAv<+LW{I_&|!S02PNN2dw;;g!ol?P7o#_V^7JBTC6fnI6_Yp(u}BTs~!j; z6nXmVL7As;jL*g>53GAJ(=wOkXM*X@r%IxK^`EX}8 zNaRKmIU9s3l@`J;><-xf;Wz|9tTkQI{@NO1iHlmy+@Tkv5iG9|*MsK} z@DE-sq7$8~Iw@|yNx>AzgcTe{X)VaW!sn>TX|sj`}2)M38epR-P-($kPczgkfP1i@mj`?A!-|jzyQNzhwogiYV9}lELKceYK&zanaSEcYp zXWq}BP1IPAGbTUr`i=3&gdW>{IuyoZ(;~xk>}Mk0ETJZuabB^>lkEQ7_pOP3cRARC zctY=*(f_fkgPuYa{SDtQMvO><-XO~#-=;ILxZI?Z0lN5`WyK5iLcS}H&vQUE2vI5*>pT`$-g*I2P4yB!ee%ho{3gk;~X z{3Ox7L0VYrj2oREO-FaER5`R;@}G)eZLPa4{m>i{wj=~>Zl8v~H4A<}`ABU;aPh?> z$*`##l?tj}y^kMcJTOfMyBfs1V;UN?mxt>W0n;+z>Gsxi-xx~ouW(7W%BHaIQD4=F zyV2iTHH$55Uf}#$yr#}S2*C5*kBaL4>Ry~e9RYc79nxzWwX)?*d)A+*EiGkwBDfE8 z*m;pD@X=qb%j)`&#`lf!JJ0W@?K<6tKLgRTP~PrMW#!&%1wF^Qsd+rM?+4uB;p=~h zS`mM2i-Fmz&$f0JXbm;Ex|41lno$fu0zdAsE|W?UwhU6W2$2VRNweD7NB0U5(zmR; zhcG_L!m|zYnb`TZB}F3p-k{Z+jBkn;L!`i7stWrDK7+(WO_qUAz`e0-fqn!?BE`7TEJ8z;pekT2vy zQqTz2zxZ7#vc2w$8<6JFx9>}Tmdq@7BNp;Ik#P8oWB)vwztyUSr8Ax?LxeRoM&gAQ zd`W#B2T%J-r3ocG%0j);cwhX+AaKUBy)*g#bu(AdOY=h^=%aw#hidhf%7DgdQ|?#q3YY`kgnyV|EK|iGL8rY=54f zYR3-EMYHZW-;^zZ_WsF=Pj%=oV$GBot>gRF#_|PJuA#U)dClxi%3sI57_N20TNm(? zgbf4G9a|ysFaWvJ&uA?!5sx3laDSAHB-!0R$gBx@kkg)vFqGOi(^aL3;(={ggTzna zKGUH5Jp#{mWeAe`8$Oapm2&2_2|+k0c1>L|wJ!D{Xn#d;qa?Y_dhoMn3gXrFuM<$S zCI=p?Jg)?PMff9$EwqDx`|SAgy z;)!x0V#1_=KV&zkldyF6Q!PfMss^7zDZJ0bbziJnmZgBMLB$(4SB&WM;S9BCQX!%R zzJaeq_w!8qpRCyt!?j65ed3~zgg7AW%blZlpeUT0{zcDn7jH1%4Kgo&*!J*TEpvk(wYYheh25iQrg|=S?K#Nk!*6jyRH@Z4C6kQ=$O) z%{MBb1(o_F!{kD&ySP@-Fy-xrfDgfl>uR`qxFUUH<|*E~q-l1Y->X4`VjFjGDdnZ% z{wJh;2$MVkmF~HYG1a5XlD(i@4f(=*mP0nO4OX3E6~ef`8TJ~WBH%m zZ{RykD6`CdBvs7>gP2bw!Wsj%uth(A`k&oXI}&u>uYH5uKeZ#rikI!+F-_ z#;I?)zl~ADaHX02;&EBDDU_cTrW;Xu9eX^r$SNK-qy_Zq*+4qMgxv+YD|sx`aSkec zUsZK1&cdH|-J30jZ3Svqj4LRdeEBX3r=e>~~JaE?3dDyre7JneyK zGM|_ZzAQ9&%z;4aNF{!Sl#u~M?P?tXp$ znJu{?L8(@vUq@+n>rJJR^FrKbtqqc&dt<`xpKj%h_W6xqufDlp^zlcVU`7t2W3?Y6 z>1K%Sn+P58=*b^7G9i}oJfUuvUaHhTf8O&i-fw^KtiJu= zOUq{M%RzZFtCKWAsqhQu_6AK%(iCq zpZo!Pw|627*pUj29+eI+&=K5$r$EmB_RAs39ran)u1z#g{0!! z^}mdoxeX}|)TUq0ODX)WAIPiE?8T)Q>?d04l@Vy|pUMPlH9n;?=r;3f-Gm(mfauj% zc6%I0dLR*u#ugC8t4}y2xh?@8k>WgZ$RH zi_#ICkY>XxR{3#5O30rY%aP6oXusvk2e8 zPJKZ6UNW<^>!Dn-!V{TI{l)Q_h;p5ZJX$uL0{Ss(Z*_xD$%xZwY147t;wjVDoQ(yE zDgz9Snyk0+1-A6wkj(Zb783jtxZLd;|Fg@I7J#czh6kdhZI;ERq!iOY*Dy?5HMPV> z9bf#GnH+ODug2|==!`a+4)S$Jpog(4lID7)BUVkDqB}**+{f%w+70P{z)R=1yYtt) zVjs=2s^2E4Vvt!OuPw}zlY;i6lsaP0G95{c(p@;57jGK$L-)w*O<3N25#htkwLBVWU z$0kJ$xP5Yz3HqFZ>*>Deq~vZDy_+2LN{3dBUr6U>id@$Nuik08tmjBiG%8F{apc^6 za&*X_;SJ4;tgnZcZu3_6%jS2zInY`R_H!@hIDhhkbK%dX=U>69BJ7~G6kOFdrSQQ}gWkaPiE`#b-;ft+`=tT*N5 zny(j*&Clq- zZ50UBSbe-H_Fq>90v0wN24F!bzp!o1JA4(O5@wrppQ-G-?j&aWwAP;dp&iHbTxTxo zjS3lso93o|P4)u==pKfNMT9Il^^H{I^P=f!YRE7@CM(YTR^f)<4_q4Q*n&63W{rME zpHJ9l(kHb)aHjLzNXTJAvn3GYik(-PJ{Z;Gf{82KsA!y`!!141fX!DsyRh-6K4BBz zpY{>XUD@TeA14;qbtBh5-^5Xm6Y4XoZUzy2Z&bmxgMQd@Dk60gdru?+h_%VO^ z$dL%cB*j9q{h-iiy%-(^_n)j| zQT22)lO|=IS((@|{P9Py`%FlAsZtX9vCqdk z&+jJu8|_lYP~EE)O@Pv$bJ6$%#0L@ZVsc$C_iKB+b7}2NdA$8mvf+2A;e+bI6`$X7 z$t0%Xc7=)0$W$f6v_X#F54-yZ8$ncXV zc!p_bgWbUGdpacv&1$Gys#rV^|9jZPhm5LIPAtiPZe2&7rOi=VLg>6)yprHK2)UCp zRK*C0m(1#;5Tr}#d$3)%>`T)!xPMv?&l~u7iNK9HqulVuBoA6aV}t%FWoisMoayD( z{H*76Z~l){fgoocFtlXND&y7?*YV0iiKM+9^STt}EHVAxJnl5;y#w>ao_#3^>I5XcX*A$<4o}O*Q0`GhE7mAvIMb_tCDBBhv{>c8rLNT z{cHN*hl!;G?;k~YT$pns+f#v}8{KyUcp{UgjtLAvl_evP)k-}#hN)*{y{UrrE8O`t zH&OjgBZ=PV`45*3&DTSyDTctG;Wo}Uj|4Ge0=iDH1(Tcho#f9CKWJ6mUF)1VJ`7j= z(P+H9iYJiY`_2%(fVtnxA%0EN68-Fx%wMrt+IJoNEX_xMIGl=qZKBsm9b>7|eRsdI z?>3O2-u;znA>?H?S5bAv!|_c6Y|o3eZv4n(>dV8X%+{s=`VXp#g3w2h6+*t!@6V=> zGL=nxyn~xeUzta^hKd;03VLi`AP^DdRmhv3oHUM(Ii(w+Q zDoQhgK>K1JW(}RU=n~CMFD8V2ND{S!4KfdqQ~_OmR%NNJ9pqq zm+8VH4p8(0C!$>w1)-pAsP8zHlqc?E4BrcJqY^8JtAhC$eip0gI|VC_1b2@}Kvd%q z<|Z^@B`-}sI(sFyAYGD`JMI2ai!wdDdJE}e@ST4AJM4KVyc1N%mo6d<%b9aNfL7Fx z6Z46@IjZ ztLNoSUH>YkfodnF;NudVtAGlA4`}zYhkWB--pjwtqE%4iz}uG&OJky6jQ+S0^RwN3 zULf-dY#4y850q1w*b!S-p|3kMu+| zib8FRt@#g^*48KXggUAvbJ$VBN~S#~F-67PT_ybw1BNBSOn^_Jtm{>_=1uN7iJm3Y zYL1@@;bS~y_b?ynzg{2Kv&oN-;xfP3$W(_mrVMMLu4P~(;psG)-Iru^ z;Y9`Q`pO!1ijbATD!TZU?Vo7)(;69L)IXSyb?$u_a>EgY1D<-0%^Vut%s8Bk$uliO z{1)k%+b0A2t!vIc5tb0B(fCSmQSb)$xw+wk+6K_f4E)*zkuW zeV2xIn`aCMPEW(=r>v&D5D|#(iOX(zpDh>N-lg8P1m><*&_1gft=7u6d%rj&bicll z;K?#|Mt=7g$(!ufXn0c*Rc`=j5oCux(;$#z);Txqlv}GsOrUbPVf^BOj3)Z{zcDPv z*4Y(C%@ff$@nyTb&WkJYUq2e0K4>^epHAm{!I0d~8D>eU%G8PyY>h9r{inwAb%(Q& zkkOa5MqHUJGB$8yEa)p(&qy9sRhF}&O)Z0x)Sh6ESJJEQ8THbr*TNd`^eiNSDR z)US6OHd7#G5T|P1HE*fKSRMOp4W3Cu^fnBWManjPnjQQu7zCZCUX7;;{m`@FR#0&* zTk|CyAsxRLoC7DcVigkI7!W;wN8~VDwe=w5+2xRl0GVVpUX8=ib&`Ns1H4Ko5bdxX z!!(zs$N}?u6N5wJO?;I-QZdfsE^q0n=Y zG0=NR&u^a0r@X(;Vfqijh}NXlP_sYI4>Lnq#^2VKu+lZy(sfK|Z4C7i+c8yX znGP+ohPQ&BKC2UmR)*Y)C}!c3JC5=N>EK`~ib+=8%pTQdX4)_T7*iaPBY8Pj+YkqT;x zp$L+Fpk&4NW;L3F6*~M7(p6$IHPM5=$~E+r;=~ice~GlC*IAtV(GI`vy+pPu}&wQi=V>K=|n)))RU_8P-jvnmZCSdK#LHwFFvOqSJy}d z^<#B@=_y_Bo0MqfqH`O*0L&hn4yMwp{4zLtrky3|X_#8e4c&Tm7+7(ZKKoSWo@`j! zPmKp{`~~@N0er8C8d7qi%P39if+X0$}R>J*q zNgs%hUpK)+8f#_$OFI$WEvJk$hR=h^F)?7>^p{@Umfb=1gQT5-sR;;cbTW(9^?tyF z&iE;*#b!YBTEa!WQ}w8bn8oB>OD- zOw-6?U_8Dc5)E~DtptG3^TGN|oF-F~6Fx${ zVi8$fvXVt|)60d|B}(BvxW1(fWuHY9tb1M(LekvXGvW%gD(cg-o8mvMHW^r^b3yj- zdp?G>F-xVVQY!9sUG|bD*d+ZI762cXGN!P=ua;))SkZ(zr5<&r`YnMv!X%0G-KltJ zNz0h15fL4=0Gzh!2bg%O=n3sPtbty*Hmz-KS&PP^VfzxO+;d=qHQzovpRy*z)*h~S zk{xJm`-P4bukKy#hQ4Vne8Gt#vQT*y(rU_Hv1VqNyObt3kyJLL>< zbzheWHeYTFL%qcHiA1`SQtb0yYX0Qa6QkmO(RuxEV8kBIvMLv9BA< z8{phC;`|iED7u!g1-Tb=)yxeIH5h2uN7{qh?o~)OZQm_`!Ve!4&czk1Tn&^Vw7n2Z zSv^5$y0oY;dxvUT?-}_$-%xcW(`~22kNs{{L_AgmEY4)(4-YZ;4i1TvWl#ip1L3ah z*J6^iiciu#4iDKs$QtXn_YOV0+QiUVuc+vIP7*^ zs_5n;LTj?$pF{jwcxR;jsb6JP3y7t`=YVPpCgr&2?DdhY|3Id>oku~|*~##>+nZLl)#59Na#XJf0qrb$_qA4)>SpoFl<61_ z0r4EZNfw+^?B(UNsELQpGa1eEc&soO#_VzFdz35TZI^lvQWlRA1e%=;yzsRs3zMwXKZ{V#q&8L{Y%*Qy`Yt=68a2x#L@$u(N4BN>Lgo-0NCp0 z-${Cx`sUoNIOKg@Ui(Kr(YG{=lv_jc#=XpwKdx=r+Xg}?wNtJ&H5K2&u z%_}I|X1^NI*)dY~DHRM_xh|m>8=^aWqL+`TpJYQRpy-^e#MDkcN9D8@gIQq|Q{2v5 ze0VZEpg8LHlnktggj3u?TiAIhrC2eX6WU2&@}RG#j;+B2*QE3!lBMU0N&_U+yG6suQ`<=r>^s#B4yyed&Ff z?!jcI>I?3?{na-LrvsR6AKZh9YI6%g1Z@W?YO_f%cVg*?{SzJ9t~RO%MV%Ih)vpIP^RGSDug#rk zrOiuU^gJg`grs3eJP~GDJ#aLEPa3J&*u~k14NQMpb?&BrqMV+uL z`1%PlXAA|-&-*^xr8Vkk+TXN<_REJ^;LCC=1i$nTf=_f$ zrzGQxzWUY}x!e>Tuyo?W92NAfuWko*9%@Ex?7Y!-1pJ1M0vTo3>S>yY(SKZclTvNV zbvc{V$(tYJoAsyvB)hc`| zfcVr<1W~AwmQ}iLHx>{=>Am!yoDNTM!$ti3`M_}waqh;wCX=#teV!0!B$=VA{QXWJ|fhr`3Xiq>S?&|EpT158>_6porh8$MC!o;FB&kZ4{0LNdaC0fw&1mG)6H1fIr~B zuyLDsGXDM-KQHs^X@AS%3O#iP>XdK`M}iP7Py`l=bPhe{%PKjx3tqrw2T>n4+k~1!M@zCa$QDvV+I~n>hbMxEz7( zGHn51vA^)X89dBiaFybYeOr9e=d|pKsEl}X8Jc7nkfg@xuoIiJ%Y)DWlb7>SW-3Sv z`m?jN+Am6Gg2-&{i!WiVKDC2Yl0|!uY+yg=v02${I5s>I2*@qmRK_q? zEW&ry-X#`-&Gs?Qd@B>f!|}A|4|q+k{ZCr@9489fK5WG5nmOgW;5-QB|J8W2a&%a~ zUPC2Ryf;YtttMS-ZfCF6`~4XB>}QFX*r6$FVG4w=}6BsL_Kr9d)wCA5=-~k zS0`6Tt0ay#&FHbA=8^@yN8(Dd0?Nj$Ld^O>XzPQO7^hAJ58d8(50|%u%Mm(BDrS#6 z?A7!}W=69!b7#!#pcV?PzxV4^CG}@i_74*~j6WFGvp*>gMw}5!Y*SuW@qd!f;XOcF ztIK$Fu}J#5b;PmMF)Ay@=sPMofa2)Sm!t3aH-g!3onye-Qr^8k(x#*7kFu!TTLwls z#Dlw{^cOT5x4Y94zv#H#CUyqSxPlj`3h8m$EQ1Gh&6<0fMoBOTvw^ttw+kiBZoZ_f zlM*BW)&}dhhgQrd3WuP9jaMz>-Ele(Ccd+8b=E1q6#cf1f(qv4FvtxflnsG`YA; zsJlnT0tq@%2T#g+P>kG9vt3#Ci3NsKuu4 zyAm9Ce<6}4U{ing^V z2P)<%@MM0VE;$kP{f^{hef050>DNrME-U!R75U~Pi$zN1Rr`KQMt1O*#oou6%OYcY zCw!6)7|7`x9aRYp&AM-&^-@{jVI1pqMyvE3#YdAoTEF8DWR4(_|D1lKH5~CCb;?UT z+pBzPbL`o{Q$M-h_rpLWx;OxB=fN)K z&N!@zTDiGozV5r^RCeEAD^vdxTE;oFX@OTr|C~lldnTK+K70I!nTe<+`l0(oBe7IF zhsP#JIc9V2%SpZhiNUGzB!>rEPhf)jD=z0!Sh#5N>iGuGZ?E<*L)xZ*Pptoi{>A_6 z`Tin8M@i;dm@-;9F1zTl7|y4sKx=C3<;()Q%aU;Fcr0x7#60f?iUUyl3n=~XT7O+3 z>0i9+UtH_|4_5p?ViC+G!L(bcXM-JrECJATay;W_l-{g{UHXd z-KtLWon8zVIt^7u?e53-bPkka4qj;ws+w5+IgnwcxO2*$&wtX0pd)M{*97VuH1S}x z0yNM!wr3UZk8ZHg*}p;?wk?o%^hQUwRk)7fL*UJzWB{-NHx;2i0I9tz!xO-4*q)TB z7=pNPuqyWJZ62HCAm&sok3{~^enRX3fP%DG={calVrje#4s7EZhzy56Ktr%JgeqzU zgg+FCgH`YgVtNlqe%ciY`1fROf2l{7EnZf(M=tK!Usf%Yd^5douT<_`A?Jhe^UMj7X7pq@xF)Po-HRm_kgJop!5H})3hOu#?rlQ zms{R9;q1X7;RrwsVtIJUqd@o3Vf8MOatRB)=73$gC0csjHSpufwEPxXBp{@~;;Nzl zu0J5HLlNKF)p7r$=I-#T#}53?_fJ~YF`e?J00q<2jdMM5dq+G+%!;@a9=<6aJygXS zE7Ps3m7r}w4hz5!yLo>2%5%}RbQq4FyJrpZ{J3R#=P(u zxO}|>Lq;%TJqG^&7Wh2%!h#$J*K07MSl1vW0x`DcSC8RXF05#~=@gWO^R($M3WOm5 zAwAo}8=`9?mqQi<8k@_+jnwFM9}8>{bx$e}E#6`e)@L(Ln;yivAqMZ`@J)RtbeS>g z%`bJ}y8K_zcouAqpzS6XdHlNAuJypIRvKHAs*=!<@O)D;ypwj!Q#8{tAB>M}bBD$G zMyJjs_jLP`O4DQkCFDewmrg7kaqodPgejc;&J?4+wLAT0DckzY07D5~k zU)bfisF55)A(hyvOdkB=%`T>4_$39=zXQ~j|Ite>t+9M#lbsFQX}}w~xZNR9=&rPY zI;vwq8V~;nM#M6@! zt=#DnL>v=sN=cC`J84zuOC1djWf4(^dTj>1kh%v`zb6!djRS|*(32Y`+&ES>X^)(r z2ZdpUWc)f-n#4KqY#_uR-CP=8N(U@~d9V$4arSQ1!(E8|@QwPC1sLqzKoV0=Gy6jj zdS56aVk%#hi;D{z-wLYhvzgM*%1z?r=jJXw@Xq5qZYe(7@a2_mzq0*ws6umDiU%A{ z@{cs=I>og7cM_2MS6x^rQHwQDuq>DFQ;x8Qvj1`*84D*m2+vV*gwi{2HRUJG3fYO{ zSJ?no_oqw8c;)D>SBHFkufD4#fME%p%(;jL!o2%Q@PULGjPI!D{k7pcwEAy3=H5S# zgt|xEUh+F1QyZXtu(7BMbKgrZB2wsvQni+!Ephhk(!rzo&OlS#4|uHZq`97cJMl%k z>x0^RO8APB(iyjny&{X~lRxX(^-g)#*-o)0tPFuZz=IBAsepVcQJgk&EzCCAc}~9f zZL<%3ldtXOtF!fyi~9;AB-CO~g3t@Io{{mmo5Fa>*3e?lkBU#k+%MEB=Li=3P1?jy z1Fzx-d`f?1HQkLSR$J=60L9Hb`(3aSazX#NLL`cj>^wEMp=%YEuu|N1*}CA{h@VDp zbAGw^l7Bi<8)Y;znR3KB_9~8Q!Q`{=7i-dB4scR1Vo;_d(u+fN($UBkv~X~N1Bxq8 zxbJ#-$+j}VEbV~rx3y*c@NmgeJfxgp?1b9e6Mth*^{X0+(l%yZ>J2oQ3t^GbBGWnr z41X%-ij#`%fKr8F@Z_t*wVL`wYRU#iNdj82z#AbDc@8lD-Rx{7^W4*vJ*9TuUSIZx zP0xy|-TgitJ?Zo{ar9egk+3eih&6}JV49rz`=6uC5i<5DTJP8~{`X!_XCCK`zmS`b z@5(RPzP@-cAy-NTRQCkGy6DRzDs3V>PcrYJUvBp1wEdcP@78?%C5jE7*>xRrWBYHU z9GAT)+bg0jXa7d|t*PZ|&hDB{Y(eY8Mwy7>v%bY+UAg2M4U2?@`+W}@r%-If@m+4C zhYzcnBk2fu(VvN)a><>wx6LcfJfpvUF8jQ%O0Zt$aVTasZi4NC2}kOe253$0i2Y0~ zi_cjqG=M(lr8#iCiXux$o=Gz`35&{UlnOh4pXyIUIH<1txnjMO-Cmt&=51&M$6#+H zcIicl*pCm)I1qqpUJRWdT~slx+{07NDpNsu88VJD}%FDhkk_X-z7oM-OJ+8;O{m-}+Ie zTOi#^kEe4&Dt!iPS7HOHbYmxg`EK}%Z@s<6#Z3QoZD=>K@EhV5I~aLka7s2-0#yw* z&DG2~-QOTh;wiAxyMVHH-gTAm3Hm@V#Uy$%42gf+l#N9aJJ4+Zhib zq-l#sG?S}3OjJ4+sQ;B?n<0-$`;m`=*EN zhAfYx9jThGKO4bpoc{?8PYvQI$M|0!Y=LB~uRYVV$1eCRMZZg(%#G|+KJyUe?=2eq zHg|Y!Pd+Z=zGL_Ymi^YRj}HlbzGxd?G?`A*t~0)N@^0cMH=K$J39UZbC~s`S1DG(5zKyn5#tqIR~F z9b=j%{^8_lEcSJw4bnemjUbBNk6yDs$Oly=*9@cPnov18Mwy5Wp5OqUHx!azDRU}s zXxuqxjZC-q8JR1#wW)ZrM|#612}?CI{Jtp*#1tTD?hXYSq{*o=oI+~;Oe7t#>&KT~ zliP(CHg)xBCV2~D*V^?IiZdKiPpD=y1gLBrB8;2~L)U^B-V*NXqOY#BES+FyEJHxA z^Qv*zMd?RbvXaRg8t=D1mSJ;pUBb_7pUaau#*Lq{1$7uHXLa)vI}-k>6>r6&-R6WM z(sj+84{|(Ng-Wddpw2mpDChXEx9dInZ<+Z@Bx@0q*oU23M15B6k{|J0ibPRx>@}bX z-keY3L@*ayfc$95@3MKX$-&R#((T2P2%{?UdzhX=R_W4X$)~PJWsUS{-Mzzv z9M}4R)f^a2KV$M(ppXlj(s@lznYfinF%2yi;JaZ& zz~LLu_#MnOG27YVyd}%dQ5Ti9K;+)H#|8Ko!;kfx6oyEbmtIe?;C+uav<=8mxFct3 z{(+%qIuL6O=i!gr+&8gC;T!= zB`%vS{O-;(z;~1&knuHwC0YbTP&zu)j<`aQp2vm!$d~S!)TC_vtbLM3yzk)f4E7yVH>=9WA(;K9??F&BJZG+`p{$++H zX`;SHx)C*Jvdtjdc~2dJ3De_5PV0`N^Ig9WK{2t$XkI?>+Y0JU6*-{WcIAtj7bdSK zr2oW-zA@cj|KJ658h*6fT!tv^+Pk=EHxlydm^~4-My5nyju#W&EGvf)@y$IuKu`AD zr8Rhge_yF+^HsMbK1~yI`@t`9Tw6}n>mg=rX0&w5{Rv}pdu2FeQkjmcgO@S7-JF}y zFk?jhv0vk~hSwstV;yf*RjiP#1oHDy?OxdH+Qp5Qc4^{B=6}34ek$aazb3(qr3r?7 z?dp9}JxyqzJbmv-2>M!An0;cgRUVwP(ZM7>1s@P8f_7oH$a^if0Z9&{< zt*O{*1aatE`DX{QIexp@=yRJXpS*!&Pa_#U`G=VG4}1^2+ULZ}28i;D_h}~xmWmr7 zYK_g7QBU6$Rlv}y2JWZ3vqW#Wrq76x&pjX0UH}U zlUqwK)t~$tADwR#{T;oYiyh#7S0QSsQ--T9JooL+DP)pZ&mQ88+~5dg7`z&2lT*kV zOG=a8`}D2na4!;mJL_y+c}uOu!m!(Xf?on&K(R_H>NTSbYjQ+0c0$O=DZa~;yQxyN zFYSfI&9xLx55z;SDVqdPh?{zX6?Z>YTD)u3l&lH8=YO zlW>8CTXM-}4d~W2?E&$)t?xz_TTth~;G8#{|HB;ZywfO=&fWA@0E#1zWRR>b<(ncY zJz5Lh&#AR=`YxwqHA_7UcYEF0%SiI8+r~cQqT59RNZMB7C1gji@^nofvUX$IX!U7D zz$MozKy8ibn}_o0;8tH0QkPp@=0|i~^Q&RTH`?Fgv~}IG9ps?#;GqM&wgoG%zQAu3 z!%vHb{1OoJO#h_D7?n9w7|*;8#=cIc1HIBSodSe33t?YpaSPTrukq`y9nfwHvAMN$ z(6yCSS<*|EzP|~8=w&szrGArXd|CMbIj|BtwYVSOO?gbHPpO;XwUcu4&C%~ePcd82 zpZvdX{??Am#gQycT8F2rRcJ6haJ~?9{c_yG+y8Vaz6Cz_<}nv+E$k&1?DJt`G4m$g z0@V=~d+3wrhB0^TsYyn--b&M>G^UqF#{Ut`?^+56OUX=xj{S$ z+n*X=%|Uc<@{ZjTyY>GeS%%oZfUcFSt(SGw9X5Fds6d>GV{Rlf<<~H3r1WR#en@~$t0%GF^+^rx|I zxqHvMqD9K-GO5U=xy4MVoJ3^A*afrZ;&i=yz-?T2HexxvX=mei$i6|3w&~KrZ|q*W z`U}oG0KW?K$@6?ev1306H`zcO-E^;TW7>4Ck#(nXkB)AXAth#_Y|d1GDv?k!gqYFp6JqJhBzzTw^(8; zf{i|0beiC*Kzg<6<%WkE;kjOSCb}PfG^Lib-hUT(AG}Ha{nB!XtzeR~^)>8ynm5CF zfXmO6goO3ww>%FqJh~Z{=Av4|SuAYdq-*5=OF#Jt7;ti=%~IQ z?_KuT%_N?;EXRO5Dh!K#gxc`pgh}}CLe&&=YLh%^NO#qP;J=nw%Lq#l8U`QZztMXy z4mQGQp9Z@H2+1UC;y%86#@cwAx^A~>EY#<3V7OP!lcmY}_csg1b^C{3eSpwmf3Jev zoL2W;nZLb@fHLn_5~jYrS}9;tZ$Q%|E?164 zX$E!7Yku7dd&vQt$(RlMOt0vjVt+z%Lrt*H%>;=r>}cHPRcxYtFO0bt+A-_9UxZud z(VhfPLS&Wp?{uZ$SJ!fn@%u96*x_+J6FD6_*Jp!vVegi)E?^-khJsxFU> z>RYGKcZDdY^0zZ}jz9FmCqwCG3#x8c{Uko^Q@`0Ckob_;@Rbk1=JB{P}IWD+}yHOMe3MvAX~6-888B()`6Z-eD}l!+ZChpo9LDGFxz0el%b6O^>TY zal`l8_{g2wk@%p}Q0;ad&@{s(bUCx7_eCkZypM-tGS+}_j`j7~)APu~X6daA7FQ+1 zF)(zl?&<)>>oldAE%%2==ktfy!hyHibm?iN6<>AHQAcA#Vo7hi%#C0Vrb~AUvj!O<;U#GbNhN6w4*RW zADl|ve7{7p5izOJzaF=r3Y+Rzzf5%7=3(6;_VzT_5q$EsoVgjbPyPS3a^~?+^?e`b zSQ>jn7hxilO0s8P5+N>RmwnB0T?S3IjM0#xY!|Zc5ryo_WF3UEq{xzWjA6*q$kb^4EL4INtPR}}-n8U@rBSE4d5wB*_|*Y|N;mY#^aIVg)n0uS4)j_} zOFS+tP#zmC7;5(Wh{*>deRJwp_Uz^!JE3a0OBAXyPkH>cDNM8!ur=tU_;Z=r>Gk*S z*W%PHL%q%FtpYkj+d8Hc9C#Nmw%}*{1WYYD;8w(o0i`_Hdx7J?n&iOLH|XKm43muH z?9y`elN!UDDO8=_z!CR`V0!t=2JXl$t!6~WLe3n~ML1`PQacU%m?k{J@JE%;Lw0Ow z$2SRUjR?z}od$Lej6k5a(3Jld^lX~t4x6AQzLBehDuyykM@%rh(frKq5(C%EI@G-( zzViZS;1O)ji=6ute24B90$xQg9a`i`(xzHNA5=XswE zxlmtU<{Qkh)(<Ylbdv};rmjhgp$GW8jBF{Pn7vx5kMQbq z(S$!Rxu|{vr~AbMf&oBNTjXDyo8@I6oWngCTUfAas}H<*kfPtERLjKuN~`wGqiCt) zCPY&i&|eqyRad_vCCCLm@EsTA@TWp}h)D)_0p0GoY76I^Y(2(U25y2C8z49Yl&6$^ za+5qK`voq%BgW>gzxzGV1|z4;naDG#b>esSOIc-GLDe`@u!DNFnyiO_UcKd#(h7Fu z$f0pdHKwgJ*>GFgIOfZTsi<1gW(e6)RTliU@MDzW)={dF>nhh&e1W}zW`37g%&?2bQ$Kk%JF|K66PbamItz6?cw?n*TasRL8xdpY&m&w^3 zpBz6q$zYZxZ#1Oi#?P&ecsYm;wg=10)e1j<{|rJMp^o&OvP8VzWR#0kCBKR8==T!e zhS&r(I~L=dw_24~rb{l(`nzr-;=RxB0a{$`gu3J50{K+2W+&9cB7@tIXbW9_JEX*lBRuWq z;9a5U-$Hqr0R#JLAND_PC1bampEB*A(QJ~)JUs8}=?ILL7hI1N0WHy;RK;W(9uVs^b&iaJ(PZVxh4W44lWi;(wn)7zVt)<6<&4$ zweL=}W=x+-t)N}or48P(T94QeS=qHRIh}Kb2b=|cKS@xt)xX@l%b~#o_KJ_|)O@SI z7~n+7b#Io?2AwUYv?NIOb8*UdxB<&KTdTm85 ztR2XPpqUYYEKV>k%POrZ_0iXOf}4?9*mNHiZzMtMpVoLMFExC)`iyVVv63ZFK(lF; zC9cCxOzJx#)5qx3QSO4)Mysh+8wEQi1qgaoq=lyt{Z6Pgth=|De(GQgK-limzegN!h)e0`^mU$YoonnwNa-1j&*>hDRif zVc+|E<>|ISB;jhlaN~)VAmcys4+5`qm_M5hh&_Df8RIl0vtVTZ&S_r6^8)$|cO_d4 zABexPaOVXP$fE`OL(deH&V>g7ki68E8?k)f$yk!ND5U&L~~D)&m>9* zT-Lj8mn>)RR|S~+4&DtW)X@}h9SmNJB z=AGpW9TsB%xXK7~v62YL1-}V~pjcQ&CshUC7-G}0c(p`_d06_*X&a;q{sEn)6Ny~M znemdJz{-A?30Z6}wO5=yjz0;ghz=Y~DG|41olDdcL~6&Cq>;xD8rT z1)Pn3`MKEB^NC;8vbeCD)v zaI?-=i|HIVrRa9*3!u;kHnrg4C!I|dKP2rjIBAdVMCF1s?%LTzN6Fg9f8Qc(BLQr;_;K3Z7z_);^B!e_QSd{i(m z#HVyB*}CcR-{6|tt}M9uC?&O(q*LrHwEPs#CQo4@GDk+Ti06#2ky$=J5c!Z$*nOa) zQBQ4EQc?=3nMH<-d_DWAuYXs|*$)>oy32IBw$p%5v+t>cE{Sq6;aS14DDg7WW*ZZT z1Y%p#x)8^F9I6qJjU3j>Dn6Euv>xDK(V#{_kcP<}A{jQ}q=$d9=F~q9k8~jZY2^I* z$!mZ`$&Eg;h^c}~w*sa%{4)*-!Htzx^mP5pC#u7hOto8k<+@Gk&JXu?usBT^)7w6k zp`b`^c;00RQO%jzL%bvDwxWD9|Q^sk{5t zJD?4CG?3o@kkr0pogaRXdbqo&a$?>?oD2mnyzLho{K=xgwUk>wO~L6q!f&+!F9Rlh zUaNVa{Watb$1R+>!yVGWpSymqo^&I5R%<^YCfjcX(*jj+9RzPhc~`~d8Zfcua%n;M z(d6O%uD=ySmuXD;S@P!+IiAjDvB7i=o=-nMC+~q`Az7H-VLA??$8HQd9~!OREHw|@ zJ=MmFo{&Een0BXVXitpbeR)L4{UzGhy!Uyqa@duGbYW2I?43T@ydJ>+w?Mc!kUtPt zG1?rK=0gQR$&)ZpOcOxSrX)S9o29FV=P7Lpf(l@qCds(&(fn=v;#eBWA@1o_$Mghc z;VqhqGi@5bQe8s)w@C_#F*VT&;0inf9MOp%cGnRtxgPaX#w8B-;QE}xsAJSz`Jd9O zXgDnJWZ6pHw&1aZTBKnWnRgxx9E-U@6Als;HDCzYzv2}cC#iUKDVii4h+_PAgxkNt zTHLvhrnQzXV*6US5JRS^$gW`w+49py(u@y!qLJ1sDK9wXZflwvS=`bE%2c#3Em56;qCeUf& z&-dNH@Ccp;B$fVmCGY`EtTcY$t^Y-%{Qp(-W2WW5s@yqG93$xa@hWY5SkD4w{29@`kL0c0DetIw)LKfI6O|2=sy_ zcy?CH6iwz6fBQ;SUZYRo?%1Hj>RG(DM#tVU&y|AbkJ*RM0FqCeHWVE2AS-7~%&!5O zu1;G9;IP|Jtf%D;&0-nDcHSiyMHt(E{OESkJ5an=$9eI@dou`fZ05PEYpy5Yaoaz{a*hZI z17&_QCZwk?o$he3EGB`ciSL@RkW^7Sg4+!L&1SE*UxuFjjX97`RnHYB*KA}EdfF-y zH41S%Js#TV?AST-g#;xgSQzIEx2HEN-pw<~%*m7V{MTqH{t-7T5@Iuy;Gzl%()#z+ z0b@V@Qn##Wy{6&1_4}6(EO~Z-h*UU2k^$P~a#j2kq$8t-fUeT|IeydVMOYUE;TVX)`7Ix|U zfdBsacvyO|34_Gy0YtUu4o2k4sOJ_(SLG9H^jpJJG1NNH@wh#+FkZ2GKH^$iihF>984E2jvyE&`}(rTCFDJdp~ z#rw10+5;asR>Y;7c~CTgB$gZ|m?YUSWpT)hh^xv@MsX@{=FsO6lawj7=Zp^5c0*vf z`ZY2RSKiO6CLCSNe{6&gMsxUz_HPYa6o7eBxv5}eCF(y(o&6hgjAU+X+@jZZnPD~S zS@H-P7t07#MV4bFeHg$w`ERQ0EcrSbxAp*H=SVlNKG5D4lIouAtA3{ZNd3S(T)cIX zdyxMpS$1Ugqh#y*5P#nn3&y6Zw2nQmu8)AnS+epirkvRyTfxDZV@GXo%3iGwh<~3g zU0gWwsrzmw%7X=dl53KvvZMbDmbMfnUBu@cnF$hcAgp!~gwYgP4$VU$Qo7)^_;F$L z-U3X=Fj_XQkA*ddiw@-6bkSAQ46v10F7Oi#?hMR=8nLQCI4U6u=Jkt2iw?X(Va%rf zq~fg5iG3VP@M0BXRIX@l!G(XBga2=Y;J@*#{&~v5Q9txl?AB(y77e)FykVqWdEF`E EAK(-IuK)l5 diff --git a/docs/README-unnamed-chunk-28-1.png b/docs/README-unnamed-chunk-28-1.png index dd7216ec48a9bceda1edcfb5d7e6c728e0d01e14..b0986dff0f727ebaaedddc86429009b9d95e4c9b 100644 GIT binary patch literal 6771 zcmeHMd011|woif}GO8#u(q0EtEMSo-2~bc(96+2vk-=61B9h3IAdo~+sG@>WMVTU^ zpacO6gdvawqwrK9AR=IZgc%w#AV>n4-U-r88}z4l&f{nk2X zpIklY>8`0}s0IRoG(Gn0Is^hK(SRMKsswQQ*fJB~BssYMh#PPQfuJA|0tAJEpujO5 zgdiCj8UidR2%+F9HUty_P&yPrf~Et+L@ahdEJ`Ta7Iy?99rz*Afg=e4$dCY%<02x` z%*@iEpmYR4&@b)j>FFeZNa-YIIujuI-$iT#7m&*}M4!8dd&Zs-C zX)s^!ew7LAP;QUMi7noD3S&*k`JDGR`=d`aq}<+tGenXwY*JMmm)4Wm#!fgRE_Y-*q(R3vIN57h1npp{=lfK6jV z=GNsUeiNi0NW`br@v5j2fi%P8me5OX3Q3->QWHHOqZiVJz47D9>iSLHo7ox7@uq>2 zr<_9f4nV+04P()LrZFvG9jdwn*}*#2ZhbP$R8Yx&-9NosNP{H1WSO{n@naY!RJ$#{ zG%Lz6ukTG_%NT7;du%&bnut0*WvpxwK`_TIu&k~Jf;IijWPZRB&)>`HK zS#13u$)8;#m`>td^P#LOm$m{1-B7*f_spASmPUk7j#G(h5i=QdTN~JAr>VMbs;^+w ztHH>Eb9|=BR4!s!<0M?9JY=j8EI%_=hVwr!Xc$pR8L~OidUdOfQn*kbb4XB-jkIK6 zv^8B>6{gZ7@F+=gS>C#j$+Bm&O8d+3PmfsBlEGf&Q=-m_g2!Y=R2*nQw0(XBz#>IoRk63Ged}_e%!g^Aqug=ir?eSZqIEvXcT7?!P z@Q%_1`Dyv=)})4%^;`!kFuEa5xALepZ4^sMjEohE>EmL~$sE)$(4~E%;KB1Ou*GZ? zR*)du@ZpqjVFJj(bzxH;?I&76A<~>KW@{b=ibc&3QzPmXy5d(?9{~kN{N>sLS1?EYIyUPzYO3~Ys@CI(JAmwdAyS=Sxd04Fe z8RCFqwu1*lA1aK6$A3Zx5}dM+qQ9-grxa(YK6|f$Dkac)v;LEwW#-%LS8&1=R_|Du z1z24|C`g?l5WFly_2nB%Din%Z;h{wcdf!q5tdMfCXxt;|-D*9- zcQ!K*aEuY09-#jhx6e1z-1sk^|1QOrRI~?&e8)HkB*PnbdUBYGJ!gF0u73cQcSYCg959pV zid*$s+ky*q{jMcVP8<-U@jWhrD`+L_xe{AK^yj zojpEJmm(>1eC`Cp&HW;B>2)vxa3A{Bj>FcpTC5iEU%U2=`=5O}GO<|S0F|6wV(Q#g%;_I{%?)Y0EjGsAF2Ycn8ul&?*vBY$7M)5TsUK#G&@m*-40<8(LaoY#UPPwS=HJXX`DMZ?*6FqjTC}k*W$Owa0(ILUH~Azxt{15%OE%W$e%LH&p25f$d9Bn{&E) zROsLV_zJPE67ZEQBtI=jz2R)wmt5)7-;xvyR6kQx@8>#9vX?WcRWh1FK3mtS^X8mH zf>dWJmEXBSxup;QcQ>Pwo$6&~e5RVHz*+It_Pwnjy@h$%>&YS=+0Kq^SL#XuDtLYV ztt@bzp~s>v5M8T+CkAX+XU3+K_Ob-&kSKF$>cyK}&4F5hTg|DMF;Bnw*7J(4AjnNA z|2|MAK1a33lu0k(RtsJa^FGxj-vTP(5~@R_#s`Rwu$0Kbk`nLVxZ`EJ>u_K0jWRsdG26=E-jr3wSeHER^ww z$#1J)Zr#AnU#b>0vrv=`5F#f-TLtcfS1i{5))f5@_urdptM5olnBp!yJxa9AyND%d z3W5pRQzs`(4*!=3bH_(!w9S^cFmU&(Klts$2^psKHq(tr6;r%A=1I(0qb)G+kBr{6 z8dP}~LOY(|c%Qo56l0r-Px%;X8}8ByvQ!$F4Ss~2GuzHPz{8rMUlietqa8DV_S3fE zJ6k~v2&E<+mFOiTcROVz&U54G13iamJI`4ekpS!#NsXg@`YOkg9#=QQ;@}gvUlNU~ zXJI`iiFO^?M?d(q&&uKbtQ~Bk+4oxUKI^2B=@-2cQ$TL{do#*&dI_lE*EE9Eo+m3FpQ%#WM1NHTU+dnF6WMj@a+Xx#0 z#_Zf-;dj2Y@~cGYPfn>DgL+;@I==#*+WNNVH?oHmFZU}^tV3X(p}o#p`O5pXz`}3V zT76B5HK5PWL6S99YmiUzE52q^9z!-=#*6Q&-CGIa`E3ny{k!3Bm;5;}>?*+CG>!JD z|MYojP5GYj`UmjN287f&wx1KYP4uY<%mk3!UP|z2OpBm2;7=)OiXHT$+v_WruJ6OO zC}`ka7Fdt=GQ6vW(+8?eVQe>ZrGRgTsOl%S;7mksGP>6DHUjzKm-vt7JKKDWH*zv} zRm0fI>&7C**E<+9rYT6vpdOm4)Yw^H7I_=UAV_l1A1GjXa!}jLOMhhlLF!+Ai#BI( z4ECe%0amBHgESwtQl`P~$V^DGpZou8Q5h}<)BCI9Tr{`#osO^Rw6%ERmYhAMjqVu( z+x&>yF)FWNQU^^$tB~?!SjBuygnaX|a5}rSEK2PDb&!z?5}V6(Rs%DwFf*J!nYU_B za$|Pj_*|MR%~oST0ZlX9ovX<;VyKY~1>Wc|zPFfPkDGf&7B|42$s*60rIAzd>b8pJ~uGC&wD>5M))M8s4K9PUrrY3DjRu5 z*q0@K9vWehc>%3AGc;M!r@v4qN1R99yzD-PauY+0A|{(E-rQ74K*f%4uVvMBi(r8g z+uQ_B=Qnp_Pe?~UZmOu)9~Kz}NJ90TITNq|N&hu@p;;$PoQg+Qc4Ud$kP18Yk_~{h zbOj5Q5y-7bfgNa(on4o!TUFWli=|!f=cF0H7AeFYYkNn0sg`KS;qhdi zvRy?|sd&B@d0lMT#6mH_dVq0gJEgpvY%I%jeG6MKg}kehAY*f;dIzV`@M7tW=c9OY zB)ps1g?CSsI2$#7^5hay>|uDUw`jwMd8NvWew97bK|JLQ@i5YNVZ(X?&sU~Y$ zgbJKzEokP)^wVXzAE~QRJ*~Aq)9>h`#Gl8h3(c(QplQ3N9;0KF>yq)0W48?m!VK2u?X4Brk)nr>pR!Y;U?f&A3Q{s2r|v9nb9XK#-y8hoB5>(= zrIDdIPIze;Pd~nLt)6uYvgR3(b1*{z#s~O}m)dN7)aP<*;3=!Q?Q)b)MmY>hl6c%c%(0#hd|-n0yQ=n#;jH^&M@5b%W`cdfG3OJdWH(qIoYEsG zx}-U-J;&iA6ecBhj2gBywXTwj zR~4F#=atO9aYwRv|8XF8|U%$(`5-MiNjj#NY_%P@gs`&HB!_9M7scX=s{{$jBBHsW2 literal 8459 zcmeHtc|6qH|Gyz*%dOm$A?YR&#+B^MRrgjBZe=Oks2C|*WQj2|Mck4oQuch1P)7FM zM3QwZ7mcxubz~X4nPHaS8S3-tJ|3U%@4G#|-_IYv-#D|p&pEI2KCjn#zMkj2UT+V6 zH#;pREG5jt!y{&V=GO~6JbZZY=Go2%Y6i$`8SusY{hY;b;G2g>mxl+*qpQoK3qGTH zkR?)5QlLtg2g$ADzL2^|P@;8_CA!g|kdT`K$ju(yw#_#Z84doB(crTL3Hp?P!ohfW zL@6jl>+(b+LDBu%9vvNB0!m4A2`QQc3MVr(^tFo%4vQ_gb*sq7rv%AUg48XE1_k-G zV+nYXN=Tq^N;sTP@$s##t!;UEe}4EtEH5WT>yo08q!Lh~Nn8E6eZh-E;&A2$1~?^1 z4k?;LD&c@C?f~GE^IHp45@;{vb>^}U56|92?i*t48xqUIBU^0ztC2-O(%hi?wcXYU zFGswu{-LSeM7|F>T-Y&pzg07l8t)h9#v4zR5u7r+1XX5pFqfMO9PH(DEGW`n-r;6q z#GK)0PHP3`Dp|_j&ChY?@L}78_VVsl+jiJ-`#Jm%FR*fIkduzQVfbyLcmIFG|CUKU zm+%d(p|-2I{^D616yP#GQVOrVsaJDB%}v!FKA&nOQiJaCsKTa&JF%+vKe{>q9b`SG zhQex4T{pP37OW((92RBSb9f+kt;X}11HR*A4*GGftqCD0{B+Np<`}=I!J%mi(`$7H z&251QQ~4;eH}K4!g&TR+oD4rOgdh$7!=(xezSKZ4tgA{`?5_L_<@Q7{dHPoFB5&A! z@3{xQj^@grVD&W}f@}A~PLvEpVbjF7kg~M8JVM*Nh0{!9UTidC5u0o23uV==d9ya; z?zut4Qw?q`Iy}XuMGc~Mn&3MgZh=gZkG(*u(>q;X>b{mGp|{4IEj(SFU$i_^AGUJhUHa)x-muxOxwDLshW97l)k_8& zLXCkI=}!*)=Y1uAP?SFWoJmpFsgM2leeJ2BK6CMXa>!{I>9M5RO7s!(^2YO2mGzmI z75SXtO77BJ;74gQG{`aMpQK`ZF?NeFR@}wT5CgWVxueBqv#M|4+Jf_e@rxj`)}=6x zV}Qnn_gnzvq_{DkAv|e#E=A0v`sUAE%#30c2%!|rQWJOJrvm!=3B!CMdzVaD(sb&I zx!Fc$wjHni2&L5nN;H*7eJYVJpVtxA1Bn72nSZjP~wdR7Ou zG01Yohxf2^w`MjZ8AvZ(JT|=SLa^0`u#Y%SHFlo)Mm!8&_Cio1_x1(+ZUAt&0DMnQ zMI^w2lSlx4_KCfs+%ecyqklqBm+FprzbR1IGafhnNcuFu)|kTfy@$bi8)oO&!c`b| zx((hL_RJ^TyJ#=fiZ4 zO{Vf<^f`a=`A7Q9HHyD`BiEHB7dEnb?%@!{P?+PC++{pYFnurWlz(ly-be&2wMdFv zYl(LWxxryDNm{g=>Uq*5E})diW7EA|%R?jQeJk6zT0h`5gxgiu8$eNOCJ~!3;IB%U z>49gMXbu?xtzrPMmaX8WJ?Lzav`~IIzh>c0pHVKJOM;YRQMK+zn1k z`{DFY-bkkv#4)C%D(pms2@}fmF^y zI7hGVIJvf+wNb#a%WAo{sPC;JdO!9u;U8=Ql}GGsRw_)XWE|{G!0PQ`T0;tDoO=QJ zVJ@*-Ci*fs6_2rsuUi9!FX5feBMtbK9Se>Wrtt1jyBvGj93t{Ux%o{NmjB!yAVbu7 zmpA+J{Oaf3CV1W*xYoeK!T|w?$RE0tMQtx44Pw5b$koT977NX^*;|s)8u%(`hrUu- zPotbll%!V-`)Hljv&)1{2nBXjsXO3JcfHw8vOpT_78qEv=x7<^A2{^>VZJH2-2~}O z)?7P(wFc;P;mvki(5YJd-=5w@Jn!KM^%+G+NX&2Xs}lLxroa7PwW7R9UFNLM<(@m% z8wqAMdeg|pxz}g^!WGN8VEc+I5G|#uT^UwL0~eMk0;OXpZZ;CeKk#~7J=!RK)o#{> zK33f&CQ^3(-SGoa>NDkCgoQTedf2fxTNxPnu=5_QTk{)A4?4_~iqK#ZRN9UEFjd$@ ziL$k(?AE~Em1SC1NJMuX)u3I&=N4!fuU%@&X${O2(8ogb1f%+?ya&mrpbnhbhrrPE9|GA`w4#}lQ_CBIF{7Ad>4%r* zN;h76LbRrDvkHSYClGVPvqE7d(a5)ygo__?PJQ%{$;{mtduUrNdYF7_a1S;a;FZF9 zM7@_=joWXCqC3VS`ug%!`3&FYc&eh8zHJ9kUt1nOGhn$X7rTYbcAODH0A)9r!YLrl zpdEFAX1aMim9My~v&<$YcUI3Q_ri?YOaVJ=?il?{*a7mX@-Tkp@vBKRAI;aAz;L7NQSp@Fn+)i- zf!E!u1S*hT21lLg88}O0)rS_cFU+&H5-vv1@AvdQ_YMp7tzi>Y0Q?u<#LCcWa*i`e zWlXU3kXV910iiu1?tns81&>MmL_XEfcdzL0oZz#l|Cf_4TGiYdwcUUB_s5aW{m^Y%B{SvCAsjncvu} z;Ha~#0S;607Fbt5lDnpMT(bp}eCD7`MUaD$9#8-NP6;ym4DuCi3&zawo9QhH(BxfSMEVM$2N3dB}ON zB7Rx1c1wXMjnSbgs{d^T)M93L^IEqrM=8q>}Gt*C^RbOpxATJtY>q% zW4!q{!8oHe9sZK-^8>!6%{@^keQby|751FjBmmv~pBm|^%j)Fz@WU!8zx$24Ia)V6 z=z_)+(CDo_G6})4vB7CUUvcchGVrRbB=A%3#>{LX+lA#r9P(M}#>I(%h!@T;m?o|sX+cfe!_5qbAXm{Zlwa**R1=DjXh9FVvh;RE);%x_r zp9v71rGyjF7Q=2{+MK17OM+`3rnA3VtP;@I3;QHa0 z>xaJQS>2z)H~AK+r(o5x$m?-od|e^+?_`{`dJLtfr?Arf-i6bBa?1T^A&(Pqz?+B< zN;Dl~z$i$_C*VcBwZg#_@AEDta?zwN@{6W(Z=oT1xxOeh=1+a#Y!B@?<3KEtckN(r z$&Xts(JP4RV#MkuKK=)rR9L5RXx*kePWQV3Hi^B=QvlE}HqRIIW8lnJCR)d88Sc+s z4uEJW7V(Y)=bTx)-~JcUd;Ax8xpa?&-hZqp;gvqyXK)9Q>#Ou9!qt0J$fwG2f$B%U z?U#i6;T@g=J?~Rbi6;#k%2;s7aw{dyTLj*|p#yC-Yha_(&FMqe|1q%$*|+LZB^!+q z9$Y^8s-m@Fay~UN34rs{s=1+a@7KLe2QKyHI`d2Db7`}(XGjy}O>emaSLhiUBR+Td z&|k4-G@l{DlxkUj5IrjujZ!<}r=5>Y{Lu#Dy-6XDL{iYYS6;bVH;R-QXUiXm0oePK zFr4au(j9?&R`4~BcMHwaZ`Vm^+VS~;jr5$qM%NOBuT)8OBrYWo`xfPdMls>1C>hh-!)MA4`8Lla}Zu4uk)8mpR z3{sg1+`8tE)%uB@XDHBF;oz#J_Qi$QJx7VW9)J4sZ~9qU>R3)cC+qzY@_E!-s-a|( zY##dq@nu`P9q8i@`K$wcaftAxtiQ=H)+ETe*xPrw#V_cZaM}4%en$$|-_xH^+Y4q; z8c&1BByp6c!&FDZ%L~JF59ls@N@e) zsydq|H|CeroeNSjI1Ja@-e7j1s@`{swrn2caVfVu_^a9;fn1Sd!IR z_>K8a^NVl&8>Vz>g765`Vgsf%vSud#3r2%I-d`JP-?PHN+p?}dE8Zr0%$U#2*M`Xb zHHHp@u{DIL^_~$GsZzQluFieC)LI{H#7n>Ln6?~4Dw-dyjKf|*?(rg&K6$#oqn3NI zYhmS=x;A-(x3f09Vet{@$GDpJ1Men_6pC$4@M=Dt1ma5!{+nbM!Hb!V4?-9u&Cnml z1v!2Wv)DMH@Ku&E63QAnwWOrsW>j!Z=t7E>p9-Y@T%I_c`gMNBQIjTLf9i-gx^G4L zyZPJKoyG6NO(K$XgkcE5bfFxpgHSX;GSu6*{qX zwaxCodU5a$zFt;#Kqp|OU^5T}K}Rh{jp5aWfWeQVYTI0P3-4Vw{&8UXcb0A^?0UEV}d(_6Cuj?!feo0NT+JEAtM-y*ePBr(~5fr6>@zQ)zTD6&M0T~6v zKrPd2BqC^ZIp{e8<;^?=f!S7OW6OL01CebRQ6amSYK^xGT6}zJ`WL*NKW`q%0>4d; zTw%(8?gz#?<6Ee^8sl#MX2-TxUxlqE;`%O*pw}j8@lW$mlyv#dU zVbV|ddKQ=?%=u;978hk@%ura}4I4c;TbY&vj#KiN;kd@Er-R?JMXdli+!3GWq<-Dq z6H#D$=gus9UB^Ku#H^;f+3$d3UZqpkpO?8oAPDBXO_-fZGwg3!qD@qE83@+Q>8Qm{ zT25_Ft!aIWOa6^0Fno8fyOQI*Po3tu-9-D%L`GVwU!|8F_ve8vmH%Rze{wMrywz{* zL(X?L{3jj1TKE5fmRp6q4)3a5ml1V&O8dQvmBy4IK7)hfLF69yRTFKkqvS`0j?jN; z8uuyf*k96DuTw!}Q)`D>{A>P-PW%n8kbV%(nP!)bD!kpyO+pu3ErXXmn5>PU9*4vY zN61N8S3bi$4+Q)D!u7R@{;doZ3awpBGJp;Q_Fd&V|9{8VK`cCwI*UJVKd?J>xRN;< zGv*^$r97tmIV_YVI*j(mNG)6`=H&EnlO)+2``GTxRvMGIyvH01pWG2pOioI7gyyYw zp79#oN}I>wQpIaa_i+<1w(!b)g+0Li!|(;*ytsCKrt(4^@M~pmdil~KI>$1T1M!M1 z6Z|gg9=GsFI@gB!38G~^zjPt_$=K9%ttoy%F}THKZTnhAjt8q~oTAyd;<>RN;yyiS zUy=1vK?S}eJLD=MN=#U->7`xmSl6w}<`+hVW_uZ?&$k9;~5;>~*OcEUbJ>vqTS zsxxT<6ZFOs-DG>{cVg`Xg_R@zx~H;05^Ooqo?7es=<|<+?SP(C3#2 zVnmx)^{lqj4~tcVNh<#NG8rwq_mXE8?E`P`f8+7Q=XL=m8*OPRh}*^fC=RQ!TYq+R zcQ1xSBs#9SQWyf&7~R)71liga!>O^v2w60dYVE1*8lVABcFls=Q&dEogVHZm{d#Sm zheV~+f~)V9dE<&oP01t%}5v=&=Z@| z_cT1R9C>wQSL8(5`Pu6r{Wjg38#Pan)Nb$`kR81H2<^O@fvj8itklbX1yAtoxE@bK z1WU7VITgu_U6F@_msM%LS{2Q&v}k60x`agWW>os6vR}LBV}_6TX^_X7s+*^3-;f@l z$5!C=owzD`&z5S}gJ?PbtH_w4 zf!rLuZrLe~mt$28MwdM$OrB49+k|Zn73i$k9bXlluYQG9nw)G=nso;ZpdU=td$PDI z55FQ4La9PFaw_Nfq$~V@!IP?h!)dRV_Zl_#w^HoSTED%u0o=Y54hS z(r2H4v2lhLt?%9k>s;A6IbmQGW=siBr+hAidVDvA`jPIH|lt z1STrZ&)&e7;!w{%O$m+7<#JKe*+h4ly!p(vW9pv1p-li(ks`ky;5UD@gPofs=_*Iw z!KeAcU!H8I_3*w?xrc<73b_+=<2j6otSpa`J^D>K*Y1==@XUZ2z@IhiG`KS%*EG9AZ_0RKgZ53OI$qYew2*TdPsZ; zJ`#Ui2*{@Pnw|;pfyXm{eSn3fDw#SQYTvuy8qZNGk?{HmZK!MU2GgjN@`{suH)&F= zD>rCj7}I!GcDeQsUq;p22~VnTIKkjnRQih@%>n73+pvYR zpBC9(SLPSoDh=M`S(X?rb5ax#&LFo_)|#c#uUYRb~M%L4z~q95H_s{tZ&OdIQ3R0^_ zyooBHfGE2BH<%}4(2XYx&q_DEx2@D4cc=fF$q z{-v94n)uf8qfvMLY;BXl^#G0J71gnf`-wqWkYtLfpmXhc4S^(&jy!>+^;lhyX{UM& zhJMKUGad(IFtv0dWe)9g@8a}$a#4T}A@FQ6qf!qj)&@arT%6`n+5WG72)iKWi@r3N z9K7mO$=Rc+Ku(-4t5~IuMs!aXB1)K^SFZ8aWSiwy&%^r3sZ z7ECquUfz8TI<)OjJfs8}cEk-r!nt!Ul42l99N|FF#aHw78XG)X-?<|HuY9z<#)0sk XlKmL{S6;0*Tw5wE6`Ef{=g%Bm@C!M(Eqhz-996>GRgW6$CN?fsh~*6OajT zP6r_gK$%4fxj0wJA1Nv9CfDI5Y4L_nGV-3aL>gmfgg5oqQe1Of$k6DYu$ zL*P&d98LoU)1H$c_j&0`^+m(>bN_2&|`tp49>&!l7acHYB|F ze_j65OhKA5<4UFE!_^O9>vL0DQdoZaBGZh18~@Swe>aG$kZrdzZ8}rZyh5ky9Q8RIu`2>C?Qhw(x)6@_Qi7zfZ8Dp@C zT?uag$QoLnybRdlq8Rc`n@wC;78X_z3PdlG?P}HxVx?y{6Atkzq`&Q_0!acjXkl#I znQ*^F;ZBogr1c2y>5dUFDL}*Ro&;S$(~vq+a}W)^qx>J_(=h~QpY2}0_gVU~jBJ{N zq-;s0#fW{jCW2XzieDJD+66Q3$l{ig@ReignU>Y8GX`(Js9GoDWb`(Aj*~KshMU|m zG2t>v6^&O(^Wva zHl!i%;S$;89=gTZ>5aiBOFA zIytx`F{%HvYEUs*R2tBLg6Zkah_}U9VTB1@Q?rm>j16OxI) z$N)7?r2Z86$l_KF-L4si(&wTPUg&_rxg%%B>VS0W?T#jCLdB*QSLGTTKF7!{$|(wU z(HzPx=a^4(U6|%FV2H=Mh>;og%3_{zoReKOSOf7m18uQp(VUepP#=iCAc=BhhyZ$n z6v3Kq-SLdWREm=7!rxZ zQ-|f*k9t#il@r)x&7L*~?Jv^za$#Hqq+*xQPuu3Tz8DMdRFDP3tZcJu7o_y6Y9E0Y zJO&GYrnN|Jcyj#)0vCDM zRV@6lIMegBuCbOo4|s?%z+YXPIapRm?$%`7Kq5W&3P3y$*D_kRx*U)$^y}(~CLX_8 zThNHb99GK@w{>36d>1Z5)SG199?plruTk4Ou`wLWtE@Q@mm?>~uwb98jRw|QPn+$G z*yv$iQgmT&O;=1Zqi>eZ-r&a2#$rsHVyC41$GmRTLZJG`$Mk3swZSRhht{T@1?G<5 z9;*uH^X9Va)xEy34BuYAM$`2qr6-m}3$n+82BvRx=ubK8^LZK`WXAD*cp+kh6XX1@_zuY9`6Wh|A8;cKhmndaXz#6!>rZP%j# zBA@bMoh&U_N ze7@U9HydJKkyW4XSCrc@2ahYVKjDZX;L4p+mILy(=ayfz#l?$9c!nO8EdGRAuH%7= z(RA;fQ)a@ac_q3fRJ)*5>9?p!kmMz_IHsbB3qI|+n-Vw?@9SKWxyD>(nS}gBT8mfR8HH!rj{3sm3#%|gS#c3;3FxwCkIlvsW!n22vv$2rpPNde^-+pe z>RU!2U~C+E9fG20f{77cOx+9T15@6^6*Y3%(voA$);0f3JuUwW9OL$Q&#*@)Ba z>UURlcYBTIeK40v1F|>^*Wg;S1`FL5?RGcM3{x0L14&~K>HuSC$CU^Eavf{M{4|lB zJ3-7S_Ln&|dSf3q*N7wKM#bJr9g=14=Ay@=JfVPi`Njm&I##zt<0ZdW3Nb{X1{IS%Mjwk|7Q>sVI{MB36A++V(Q|>kE|X zIyXHmu=XnJ-9(Qg7V;anuM~j~oMO9o89ssSGS5xmn!m}~o_Ae(j4NFt0~9-TqhGiU z2nA-u39p*xa1e;a_~bkvh9R$4BFne&10V_PAET|xaH6|DR#{>kq+(fHINj8_prN!? z)(_8f5Q6_|Z($dHE`qti-dw}6k5Z-QmZTS!GWI-qmgOp}aLswz*ClSo&`gT_^_axs z^o>(&8NI3oYm2e>U&&Xp9mf%(V(?qm7N;h4#XPXi+)>jIqdm+g4dRh3D~Z+PkTmbn zo~3|`CCl-~`BEsT4%l>?5bU4{{R9y9`{^5agGz946&E#9RjAUivs0VU^yNK_wI2)n z^Rf5fXyIJ5-lIUdvBi~{nqSc!1Nn~5&MkxHSVa zG%?ZInB6%*SeoN>nMM`zx$qQV6>o$i=6v03`Vp$yAN>UKGZ3Y>GYdAuh?oR0`FHBg zAwM8zI+iiSN0=~_uL2Mu5zeN6Z=cP4( zJ4DNadmrB7$E<0Tq89CV%MX<3x{1Rtd(?AjKKt|7i|#eNHHipH#x^!P z1s3e^-# zh%coiY>ySG_v^w@2Dmc5>}ex4A1@YL^e18%tLBXQMe~pdjZ0kd@>Ra%4FzH!`tU@>e`S8m#=_40G|*9 z9$t1_dh_MvpE>|G*io^QEPxRBPq!`q8s^_bM(^JAG(Y8ChpSD#mR@M8YWfGTd=upx zMI5$>HjjD;r1H~)!#J<~#*erZy451!AR-?pb}^1Owe_+A2hn{aF5J0-3uwdn>tFNY zgxTw-Z7NvLizLJRf5Do6G*~?^xu?L|mRFhHv+@b6pm+N}QEq>jL}QrBgZp_z#r*8F zdC$A--{$=vJp~Z?KZfT2s>pR;66VM;&W=59idA(vRXRIN%3iUu(tplodR8`qgmj(FxwH3nS;}1({e+vxi7!P=^ zo{n>gl9D^)kZ1nq2C=d*>P{Jw`D;v|Xz<tw7L6wOTO0X9tj z$;AvyBF+EsOivVP(w3a9>E>sBs3LEWBxY376NTyu(_)h+RIWC7(u^eUcu>VxHqZ6W zOrYb&;CfM!hMyK}Ei1EUR#KNqLj%;aL{YIJkwdfwD?+eW%vnH-VjHTScu;+Gg zR9W@N!NW?|%Cr7pXxx2L;#>EG#U9CyXX?<;9g#%kS*Qz4-iJ|B$G-i%W`DI|+WO~K zx6FyoqnTNI2O$!1V-+cd)a@6YVLz>C)5pD*{zkV0HlrR`NHgJBkn%DcJ*M32nxMRhfn1@ z=e360y@W(!gmci3j6Kyke6VczoobxseA==OmJXOew!O}9tHZgm{>Jd1NGspk<3E!9 zhIxpdn{m)Y8{(ic%?d9FUCjsV(7_gr&04CU8rn`JcoTxACVuHS1J~h`Xc?)Ow_P8Q z@6qaxoF|8%aa9f2Rev0OuBM2`aSrn9v`_iW8cFwOMw|+I*lpfY$-81QE0D)x(j+DkSTPvf? z&(Jbr(FRT~x>e@@`M+6=43-GjjvQ{xCg)n@ z3M+@m9jGMqt!dY6&xt4EP1!EbCMcv!X}^9mj+mOL*6j5M_e=!eKLpJqK2pAFn1-)Q z3L^OtE16lRCfv$^$yp-%3A(?Y6;lpbtiCuwOCh0^Le@QGQ?M7ZWP4~a4O_y@(*c@; zZ_iG=&&+mefI()ilSG0gMhdMyx-R;*tnI6AN6{y~Ps+D9nm6g zv_)BtZqI~_4BQp?U%Q4RYMwP-(Q`y|B?47XKU6I zLmm~p*bIYA2a{gSY$OfMc{N^AVmqsV6e1chXg^UsVf&V4G~J_r`_+tu?CxoxL3yndo+Z0rNGt^J-b0&&8aLX;|H>iOla$ zC+o_wX8ya+RBX}%Pm^SzN}o#coCZ`B>cT8#xJRXSO_SwZGJq$e`hLp9Q*>QjS&?~h z9E6Os&1abDU9wr=baMx(e`{}Mjy%|rOh$6#&j;CgpWljHd@*dl{P8MjW#$}liXB?b z2TqX+IeW^nsQ=-1h$%bW49N*G&E4MLU05iQk?K&Ll=9S{)Ef9l%}B!^BF5EYbWM+Bm~7 zSJwum{5VshMk=}6>q1)Rh2_e`T2wNg9Pmn^8E0Ci&{g-W%|c)xv$Sp#)3)-vXT-%V zaj+=hQ3^iTp;Iu-)sn^NSzVMM!xA&`|MOIkkRiy;*}Do`^YUq3qiosY z>X~gl9Q}S~rq=~dVQPT80jHli9+k3V(sqVDcX(2@Gt@}Wxi zc>?+%eZa0;{8fx6#yxaX5e)N4`(f=n(e)Do%qLOWD66%MAZ64mhRnT{gywlwS68di z;dztUpCtv|8xB3abYFBj)^i%;+YNSxT~b3x1ATPp&RXpQ5YHC=peCP*O)W^!R)*As zTg%Hdoe&cy?29}b<5yga&z2A?^r|^V+6O>z7{51^KoqE$vqp2h&W3o#ez}+UnN&Pe zzuMP~ZF1)HpKqEfZ#IT%A0t`55U*c!=(=3kSK89srrnl32joJtQpi~Ka)4nA%80%z zwkQDX%sF~ez@Xx)8GT^Esms6VsW3P5E~zrqE3Vgkg`EI_z&YIi0(g?^TzwWGb?(*G zOUd||?EFN1`Pp}ufJf&zQFvPA$T7o;t4HbYX@f8mtqKSa0W%PXHPzxoC0rLxRac_% zl8KBKwJhA@ffI6oi()OlAa?`eHSFzi)(J37HR*z+k*nXr73{u`tUmix{9YWT{#vNA zkP9B4;4h5A`Yb-ioXd-Qx#}{t74j~kaS(`*=dKWT^~5N>#o~RfZ(A88NJW&25ri91qh7Pi?5`}e~#xxIt!n&hA4ckEs3c2j5g@&UuwzpamsW290r(D2AC3s-sP(a zK3XtYAyE(CJ3ZE?b7-U`schls-|xMA-~azN zT%8@(Ds54ck&#*Jc=VT(GBUDc;E-7(3t;+~;;q1iXDgA&#cYNtgX$grDxJ5wZSsA!7!j%Z6>TXGgwLiXsbtUEgjfu>A+c1 zE1}m)B(KrvmIn{$nJ{{0FufMoGo^HZWA&iZC3J}d=utw?l(GXb&~6pCa_WmSpsOPC z=;GuOaIdlB4@ib<5&hY2CE<9SMT zXJt7x5*&Jb!;nZK`FySo`f{o#vSitrv>3DEibVMkwu^3rosPU2s#iW#=_yl(038Vg zUnk3{Bq?fw)r{qK2d;4kgUsNkNXb+X`5L}g z<3B$#v$EohUBmdifT^?LgWb-lkfvQ;Q|shJxV-lubJFg>nQz>>J^s7AD)HNr=ZDJ1 zGpu<{zMvxt`VgCkB|Y-WnqM0<;lgNt_zdn1`1*o#Y6#7fs)}V{0BGb{tc%*z)x@=e z;S2|tV-5_-*Tt@Sd@rQ5c|xYUxkwUsN2n~Ai#ajBfN7YN|1_helT9rO!7V%^;He7d z%^;&Qgg(U#$#>SD^6RLZa)(_2sBmK{=VUfj z{W~#G!#Mhj8@bKl1HEzezYn?HH@oR&Bnor{vPec|jT-cKH=H@pSxnVP7E?ziIET1r ztCD&R>5GYOZ+-&DZp`hkc1UEh=ie`ro7+4(m2)0yY-hfI;Fa?E-PL2a@KvSbhjn2Z zTQ0qoL#F~_e^%qdz+(l!oWw^<>UL*bA-q{x(U78cl(o;i0!%=5r7D5;_HP~(04O%)9p4}|BbT*YqvdQ^6`delX*P_{>-t%R^C)a~9 z5538KQIz2U8^@5{{L$ZzuPEjC#qr=jlNZU)z3;e07a}dU?YuuUUsS4n?fN?-1-@AC zew*T;QCSJg31-6fgEkt>IyS%Mjp-*83;qw(d}}E)I@52@{I5U8LehcLb2L zt5S1IIrLwlffGf49SXu{T?^pr>x4>zAXhlrv}tM3?h4l}Z&F@m>%u0ks0kNTG)tSIlzxe)7rkcM6tjh!jQa`@&+&?35k~sJpnW)T0i@P5;Oo2x z=(0@_=u?VLdcDPw3=|lUn%7Ny70EJe_QuOq?eEC-u}u5R{;c%4%@`IZX42*|Z0Js_ z7IWL1?Gi}}XnJ5vp4tvl&LIm?N|*%SjV`!K5eCVP7}X6QP*Df61jDwx`prV}bM1Fm zoltNkznR}u-|PRZtZ6dI3opT`+;rf`tMo1`TXRQKzc_o5GaVWFk+)*sN_@zf4xEsn zdsc(79KBPW%9^mF$bIKutEn5GyFKJP(WXm^LKu%QBM*bra-w18Ne@Cl;l{>ScYq*aBV6-2|dE zcBu#k^O{4NvTY(}{qM9XZi8m4Ddmv6d8VSXsq#Qtp9{6VR@pcd9%nKV_FTIWiG7;g{EC#1ytocT zneGxMERVEi+s!s-J*9Men)Jw0`n~42#2Wx5#i~Q_OSw29`&+;gxmX~1~|C;%ho z>=40hw#z6ETN#S6ANdm?V8dyf;q0d2)XihH7~4LebV$D3#P~c;13lxIOI?4B!7|;=o6?q^jk8s;dzi5$t@|$-BP7FQHxVIvh zhtdr%h|EsTPINi~Ih|J$4%B<2`@$uz4K@$`%DK$+z+-!U{K?qmE@oh~bqhCKS3h!% zW_RdflnHDBp0I`3Zacy1G=LUT>k*TK?#DnBR2uekORvAmN1;&$p>E5?c;9P@)(!gw zK^Y%Go;cp8Ry$imo@b9;s z3v3C*oj5G@`w8}!ykiwCAQ`_vZxp7!nb>)2u;EQ?gq$kFNI~?xt1|*HO*?kLW85u` z|7eg{`bL-g?o*%hC|(#rSoZ4uQ6m!ejn+npV;}mHJ8T1*v?gn%zPp~=o%}LfyXV5> zA#NFQ)tg~Tjm^7$!>ne>FVyksA?DNKj!?)S1G_s0zMij`W!tVb-YteePNM~l!OLq zFvsj)5gU-{$Ft;A{L1U|UCDnIG#9~b{+V2+uqt&fjjT@>MOM?LB`t6%JFl@EtkbwH zUz`>Q43WVAL_9=)^Tr0}W1oqI><^qc_q32uovPUMncJ%^b=c)4(M+2O9{N%QB`*4;`x^#8;UC97OtW9jc0_(YGEFra{Z>`)O19&{ZivT z;hDmXbOk94iD;gRlbngxgS9vi0q1(ci#my8$3<_yuV(47?pV{QqKX8Y;p#o1pE288 zShlU^O0Uc~A(P*)sTF>B5$QUY2CirqZq_l6tE1z#x3uVH6x>ZGMXVgZ2iW10@H7dH zYhx!t186vd<87t~fWRnMjBjumvCwLik8z#Bwmswuswyh&HUjm{ldO?EbY^-aLlydc zE3jIVV)WBAK2e4}IENL1$=+Tji}H%~z45BF8v2k44Fn)YyvdM&t%icw)Zz|cx~y>t zq;b3!wN}&<8u%exhimR~UmNhFAXj!cxRn=cD@ho}Fq_rNBJ!o)hXZto-`)Ml^o3rd zzs<=~n1-<9bw}n6>bt)&>|N&GS3ru&c89$1bzZZ1K^>SpZ)+lfW4A;4l7Nflw!WkS z@MYWq%2!2N=>2_xUR4CVkH2#ry7^u1A4;0t)eyi*ZG2RO9gNVxY_-F&>4zf_PuP#I zitF-`Y+FRQ@kCOaB0=XHb?mD3Dhd;}cOs?icCF4|#T{=mcicvt&(Oq%&BMKcwFC}S zh_O@}XJ4dP_iX*jHe`0T4vr_c(j9>14KVK_?$HqoTR6ppI|XS?dVxo)IBt8P0ZvNl z#w9d2GG%!3K^A40wR1n{2x=w3*-%q ztkH3lkf1}-JR?mes}6b_Pxx!S{?>#44=~hZg{!@!$>pERTsfGq#-ddZl$C`% z+}AnhoNF6ndtZ27a&gB+g&JRXTE*vqD9bZ zfk+_;HOreA>S#kNndC z*EYMFJJY(!ct8Nx!uj*dEqZ!?*B+|>IOW?o`g6`$Fxc5-j|AN-ZmskD$2D{nFA+}@ z3xNh5h=)*mlfi2s9*PoS0kt^K$e(~Q<42RiB$XrE0r#7k1bX*twF3;--?^kc`Q1`_ z%Q*l!JDCK;&rLYc3XReOShrKNr*#W;qoln%B zh61KIN181P%P3yi1yDYj>)f5b@~TOy<xl=-*WKns3C)Kbl5CQf9Le{Y z`K<+q>AD1vIsQibj*b51LnI5q)Aj|hSwN3nO&L6V3?Q9^d&l=%!HskX5B$kv$_-C+ z0!R#R7f=8qVg?CSP{sOKp6LY2P!F1)aKdU6(v&t@;I~tKS8x zHkU?t5B+rqVQ1jt`$p21$wlQ&#>=*U?>Xz#jO_K5uErvreN!fgshZF7yxV88m(FCw zs6I(q!*|uq6Q|4h$FNC3FZQa}FN1CFfUD|b0$%iNRPM(l={hEP_SaW=h26Y1Ln3FC zkjpOjd+s-DJ{%XobU4X8MVK*A&kK6g0Oju3%YEj<-<106i;jK`!R?LlT@}kWn;&~k z%qWYFDiK-TnC2k2{n31R>`wXr$=m6yA@)l(puTd~Q~s|!b&gM?a!C7c4sll&7Dvr} z`iKxC1&GbB;w_iDpaH5khf=c(ADi!}SWcN`iC~Qn(Eal0>iM}d2N%O48|q`Y&)VA) z1LJ+2WY^WOLb9gCPxe`St8Z#w$^40%W~9(enbV&fy>fz>+lO7iMpr)zT4`TqzRIa| zdU6V?+pP*LS0#;mbTTg6`|zu5&SO5T5!jbDCo}AE_pM&el9(mJVxlEBu%anJ^=?MS z;2iVnDXEdHAC8BafAH!taC)m-^dkp7@JAK;vF< zpa_o`Bn%UHuc0<1@A5!n>YHT7V~~S4Z|vAF-uzy-@z!dBHrt5V?)=_npUWls^6cZh zxJutN&+3rk+a8!#R$B^y#k|%yT$_I z=0877?+(&I=AVKR7Tbou(~sWnZj7D{BC=FlIMZWNSx$=4`PysOE*#y0NLr`G6;$L5 z(zi3I?QEPo=Zx;HLhCgG9ZhFOXDYy9la#K-UHG}?qhf_J*E0)oo-iIc`ORQp-h_SyX6qzdJ zf*{_!?=w$?J|!^`3RAzNH=MfM^Znx3qtBCN`3~@VbJ4RGpTq>Xkj=N48L8cf_P~t7 zhniKYHQwmY8U^p6HDcErwA3q0Jz>Z+&y`k_)>Wnar)t({EKWE+RgvSx#=hk0)V3*d zPA;j0To$>)e28Kc;8=m`$nsN#{u?DRF{WSZRWnN2{#y%=HO+uYt<4Mb7Gt{4*R!hq zD%qeGZyuwiY(5xe`mG-KfjIM@e`E09-@@W)*?99i4dAg+eM`a>ZYPh|Tohj;l3bU} z6VmujLzL7xE3X%a3D1RAK4;=Zf&FpEfnwDzYlH}m$~XinT-$A3(kYL&_FgVfAAm7C z50eeoSUfdhI@V^lOmFbY8oZtoU5|IU33<9WZ4NV(>${H6%7ws%>|u3&T_K2Ydfs|1 zLmhk>=(G!bc`2W(qX370-pZXg#8(3mwlPaspbS~yHHW}vbYOIFPYt7KZ=7jL#BNEV zV5V)3&C#k#qJD(xhH3yEX{@1X4-B~B~>@`yy`+<&p@D5d8wq{o4B|FxE0#U zvX*y2^rR7Wq967Q0#6G$Sfb9Y2ks=s8jFR)skY$@Ll`Qgb4OV5=zZCYXV7o=530sn zG;#zXw$oM*EFOwDt=a0qxoHIo#5$eSxJH!!XFaBeT>iJOZWAY*a7zb8=S4XyZKdo% z8-=d<&aB8%x0xXYI|MUpHmAB3G&SiqQNu59_no;{CbGvBxThHn5VRMb^!2WzDh%6Y zc+3n|(I_~mdV#4+;o-ttomn$bDyFWs2eH0s?u4sM`T&&ON{VJ(o@Pt%$D2T9w zS#p0}&^^_zONDOY(NO`r%ak$Snl(JvHji(UbhkYey$~1FJSZ)038?HYxx(lzV|7lJ zcTHzqi{A}$p#@tDzLc_aJ9`JB=O{(iZY#^!nji!#uCYL*3nJ(-OU6LB_yZU{7lRyY zQAOWW6Oti|icYCv6V2ZO>y>$(Q5=Y!LywzHD~!SDrBgN&Wtb%EG0$9GdZ2 z2^TCgnhe!d9~v4x%l$oeKT2eJcJkJ(`7bA)*o^Z}=Ug;HiL%Eees%1yE8xqvQfJ1V zMEqQ;I&#hJm5#`a;YI^rCSlyKZMbh;{MJCnm1p**0rGCZ*N* zs>Jh_-D%1x#qxbO@mckj%+xn;XJUkA#~b)bw7zPy`I~qRj+OE_vDvnK6Gs+0lN2)h zYg|WKO*y1b&;&SlL3uLXXFc+yIRp;F9)yhw_^g$|R!7g{Awg;Bto&a;J_xQ^1Z?Xm z*H;%2Lt|}h^~?kNTdRd0U`9SpY4Xt6(CD4}^e5!7#B2t38nWt6-EwfNYC)jz_jvU1 z`>01{oj*CG`XmAtE(7}$ISUuK6x`PRa0CCX$a&w1u~263jjjMF4%d#?Vo-+s^x1l= zcF4u+bKYQQc?)vatMe-Dg+aoBW3LS=)d^!#ape2zINa$#^iQtC2{SBwsppQB`h|y1 zz9gM*jP?rZH8O7&r@jAmAhY7T|y6J-yf_{oE;Y=%cGh=|>li McFw<49uBQAD2CS1ZHW7TB=$)W@`-{aAz#B^dR~nW^s}2oq ze)y1(s7y$7B4EJ>5THZQ`{^$lfkqpwtD|9^XoN%>0ZRi-bUbjSE$&Od3M6Lr*1hD* zz#x!8|3dU`ME=3Ru#Z7c>x^k&=Ijv7v0FPG?Nbd86|<}pvW=RL zsqLO(3Q*XZvWWc3DlqZd8sBqynwIAi2xSUJvauf?n^#+a4Z~oX-w&U7lvag$n%ohQ z#ILe{jKq!{GM$SHGV@3;K`IWU#d*V%0^jtX)r1zEtQZR>&hp(o?5Pn(kUqR~LVhRng zQaN+*tG?8{2EVwJA0}%}kwj*MaPIp#zH2zlR;u{H$X2bixWy_@X>w;YE8CFsvoH=N z$-W66T|y&IY22^2dXB(j#|B)B*^TfDzlo_J8?DcHasIu9)bx4Aw$u7=h-#dDSN}J~ z`bS-?QDs;J{!PXFBZYraoBx0$vqs$|*Sf8SM>Z_4)Kwt_7Mrxv&ItegB+3HZ{UBYx z%mdNr>EDVFcHxH>|I(BsFvKsUthA!eF#iL;kQ`nDr-=Q96x*_6$XVhSl7#rbf)ovG zRXFmXBffs$DG3(K`}D^ZL1mS2v2i$rQ^IfGki!7aPMa*92T6C(F+9p0o^Yu%ekR-t z-Wucm(pa=6&S3}F&xaaxJF2iA!z8AX&l0GVvpl1Xhr6Frg)~Q@S^KUBLgQH>Jf))x za3LP3II`_4tCBceAD`05&%&s(0%>2Ef@%rwN^Q*N9kPcZEOUnt>TJF+eH2>W~v1*!U+lK7RNFMJiESA3b zORtGRKkip~b#&mUL*77~mK?b16`ZEX+w_%w*9P_T4{|xl2-rifd)Q&P5HIwLus22h zN5nLzmA?wbCw*&v(T9T2M#Q3XHSjesTm5e65Xx|1w%&{8lOmKQ9G0&fR5%gS)@C#21FRbc8hr!)*JP)To5wN`GZn| zD>l4`Ro^kld;63xsba9Jg-4y)lZ);G zMQSA1>{<8Bu>ju-GdnZLN`{JW5PXMx#m#ND6}X0$2&%9k&n>%$5+@yQ5p9J<5gSOQ zZC^-cu2W6RoIB$@nI6XV*>~G73OSYB6ow*Eb^DO>z^~%a>jy}qA{EOr6+PSjVlJ|v zs^VQa5IN`TYBKg^I-`WWo=Xun$6i{z!=x5TCpg|OLmT>$)jL$03Z}mBP)c|m^ zQ^;GVp$)A$J{~dhq+l&6oaGNsbos?oMe1h@26(+^&rZp|Cu#7TFASR`)_?kl?k4gm ztBZ;`U!m$3Eid=>@1H=ZAKw&ni6Iv{GtEC@?Q2KD2YcpkqE&qCTqJWErDu%r9&Q`Y z%Ty5( z^5$W4zJ+|MniC={9F?gY?is4-3ia2-KUOVH&4Fc$wRB;c+aw>%sB70$ONoZdslv7K zo(x-NE>A%5lnO%mMt$np1s9gCJu$1|fg^AQq!76;21MfiRWjM$M?XUUWq zw$FX>%)(UHQMWo~zO&n)3>X+OL|KG7>*^3_*TNgG!FdOLHt4c!B<391DNN>8f7l_m z?$&YqWXd22j{X|kNMAdm8^Si^Y)aOPO2h}%d8VV}%fz@#4<}5Y?dj2N^UX0*;;~Motx?vLAUtq_+(T=|JRVw3D9&Tzw~KvJOHc>rejJlxQ=D!=Lvvtw$H{5cZ7N zW&tjg-1(L7r*;QJNSo>(t1xZ)b2Ed$Lw_>D&*5B@tpGib8O9djpb9`g}=uK^N;m6 zTd;Wz-vc_ezq{VV3jNjhH=zafCkTL|D`WRBf8~T zaY@W~Ga{=qjVGe5%P_idpgXi^+%#0^zdxtYvK3brb|rNy>Oi~-Npqm^K2I+W}gD&iiFoHggA)} zsASICy1p-!=T5!fslVW@5j(=XGOzIBghfBt9BHzx$)uVdAaT7FN6(DBSWB|I_j74u z3N)32DDL~+fr-6=+EP}5Q=8Ao=o}z1!G&)2(dD@O;WQqe$(;d|lZs|dL(4nRMc*X{ z;qlgX2dc3JF&g@Do6+-l5U2NS3jzRHl@8Kbm4)!gL7cY-KZJyelkRO(1%SVZRs|nx zoQR$Y?Af-0CKr96s>z!uon=~kiFzG*t7Wm5b}exKfxme5s1fOF*)%}$)RCvUq;hd- zAXUHdo8P$gx7QPg-oJc>8C~rD{&u0u1wM#I|DyZJs0Jde7Ge?Th0=*H9Bu6UWVf|& zmF#;@#PA$`=Wi<%)^=KRW*;YFvSsJbs1OJ5e3@A!{Y9nsBEB%!E-w;UUec zQm|>d^7a51wl3EGw)j|GN&gUCbijaZwe-S^76M5C$`>+vojmfM_ITqSIv^>lzJUlU znLL$jGLzgkO1(9AUI;QTt`ZU^5?xmYWn?tf6wjT&3Wh~{xIlPn$3H;bBef0Z6(^L z3Hif-pa-3x0V0VK1LHgC%)f#g4h;!Q5m5Ucfo8Q7%x9f;>h5qKlpys;-}o}dpW~JP zeR}l(5;}YUu=WS#A#S6&d+~5>Iu41_D%>*8BPmucuLg6=LA>s#mwKwI3FdDxM`GA! z|3Weqk)5I(+hvV}s_!=8`4WdCw5pTTP%0qK?X4_%W z8W6O8iY^#Vbq8y}n}?S0It<`ku!Jhr5v~69vI}(B5l}NAO`C#QT}dCDFOFCv=5n)^molfFwE)v#mS+6;8zFYxrQl7({YE5b!f>F34IKB=>!J zc2Ex8Yj0}unN=sAf@Nb{HUA^sf2(n2E)~6g_jA;Zi#q~u==6Vjados~VjmLvMH`HD z1Ywoe5jR8+VNi2=LqB|TB7rM%z%hdcNTQ*C&>U)Thm*ibN}3!(OQ1L z7&D@HRWytm+&sUorV2JM$LR{I#*hi04&KxW-vd&VU;eeU93lHs;;3r)`rJ@QiO5sC zVNudMp+8*-C9b#BUeiy`cssFpGc3kJaG}kUY*&I@*kH9}{OK_jOR6vP`Sq{Fz@N~A0o9U#4aug`nplm8jqpAW$4*9@Er9&rQBdC~FwSA3W4AAJSJ z=xQ`aD5f^?MkkH9B-T2BadH!`BP^ylh0E{?c4`y^-J!^JcVN9OEklL|^MBvZ@VeV>!MVGw?N~T3BnlhY$(pD}^NvJ?5 zU9xQ@;k6ZYIUi@o`vhKMBtt?*`86rO2=4YCFBR|f}JR*j)QPdA$1-7Zw}J4 zQC%IHRv%Npaj~QM;1E;I-XpY}vVF++1#9x*N$s^#q>H`N7ItE1256}OplLJ^~WtplFGMhd#neY%DFZ7or=9`~#;*@!#J|>)hX{ zcNQ-0-8*Tq!>52L^CZY(*biA@@oOF0+ZDssZ}B4VnY%-h(P3RnB7IBm<`h#$a}^Ul zsGk)_+PT{vok=Qn3 zRD*4e2Pyk&Mjvl)b0klxtU?LjTFkQ~k;@nvfWHzCm`@u)FcX$_)w10?OSE0WPl zvoqrqJ21((YsXOL^O3Ym)551%1ehP^h^`;(;sA9j30R%wZR1J|F{Sx|O~-gC*b=Di z85z~^+iN(YI~S&+U(W_zbAE2Y?vu}`irjuEO=Z8(Kj*w9D&C>F2&zLN$NSFTdagzF zF3&1YM!PYEv~VSZ70qZK_k^}w{NYE`QTon}hi*oz^&~qqPW5)>yI3Lky;oq>&@3}R zke5Tf5UDMqT?deMZ|#n1#e$i@-ZVgY7@i^u*x?S(=gNB;lkuVtA=mc6KvQ;}5H080Ycd?3)&A2d}!nzYEqC%xl}@TG2c}Pj_agbhd=zLB(sPMXrS3 z;o>wX1z=F^UhVelXk;&tDhXmMrS*KEqi1RFlbM%K?ss*^1#=K(Nt?`{rEUEWB4;Ac zA-XFFo87sDnZ={T7EYohX)KggASU+4qT$sFbr=HUhH9~bf+GfQyJTMkq|vxH=k9|7i~ME^jsgW1W!g4Ke(`{`()Z@ z;qD$50T+u&i6OWInK#;~xrIRx3Ch#?Ye~Fy$4}!xDcuB=GJ1S!R@G2zZTsjoZq3j9 z@?3JA@nChibC&8Z24xyJOLgG|w{6or$HZyjPpfc%<$d=-B=nX;p3OwZ$r-@eH*@9N zX1L-M`kM9Aj8ifz#H5if&dCytOmi;(PoPO>1rh2i%F>u0HIGEE$Qw^Ozr3Ln&kSq> z3ie|m4$>?FF+fTLw;YFMo55_c11?_}@}1PsZt^MAAzA45ryN#5IPG zU;YnYn=L)CD!(sWksqo^^>;xXX#R5Ka)9%s5vYpZsvGFhP3Xka2pS9)>miG4^=@B?45|<#+PIdHmhUSC`#u_zi#%KGN2t{=TW} zZwOTqc|`Sax0Pq$Xd`!2Z9(jl%5^R|-?p~bm zGU`?sQ3OO*Os2=BuZPO>Y};@WLxfDugSI~?+8RDzVOMA1^zjtGF7GE?yLVe>VE+nWlz zQeGHK*V|ynY{p9abiJ2NvHkOK^vrjVSThYIQy#{r#uQ&ReCNQ%9#uN&oU*nt5re^4 zp@J(+RIT#xAxI=c2AF&$OC6rwiNYV>-Ju*l0dcf@vC%)6PHmku7owNJV!N&y04B=j zoP9f86|k>zHr}m)W2IwRv1jpp7CRxeb`L64`6*&GzB#M{8?A15uqMlO!a%J4IpKhR z8tZG~*`Q46$xxB7u<{!>EldsU0UTK@j?e4TlL0h2`ol7I`Iy8hO`A&%xf9j@d_S{}_89R>`niq9dMcm zR$saT`(CRH#-hvkIad31;rHsjpLb{azQRGIfoR>0dnNya0Qdqc*G>) zlz0dm{NC;Uwb~Gc<}LRpKMzY4$i27F-11)5q&`fXUw~D6<(A!EuptwQdXI^F%O~K% zvr&Hk(@+RFHdQ&h{4(5i=XXY}7ii610$DmBR?^3WY83xXVSzlt9ecbRu3hjTvu9NeT&F89Q z#!i{3i~)!X|EkATNQEOf>*d@2ML+1Z{Moqu z15eY)q(l$Dhko3(LuXLi07a{-@u z@23K*zMHMlYx+IP#+9WtDds(I_)KLgod3@a%~Yy|M?1jTfp7jtH-8t1-62WN6e3Yo z52sEx-Xtb_X{OtSuDUJW*f4ol+L@OZ7s5+y2y-cuakQ+Q<&Y+|tj(m$;P!-7xlk0I z7hcKI+Az^Bqr(OB8Z{S2hYY?_@{ulzg|UG%vlYilLcdZq&ECiNz7)-ElDZTi+(JpqaEPTOkuDP~N8M~j6 z%}-Ly2(x%-!?{*3=cj}?eYY0)gLp)6nQi)79y_j+aK7*gZ^ye!__7i!7&G6(NMSfm zNTQ`hmKa%_s?VO^gUYE&HE%g;>lxu_SINX7Nm{{tS~~My zZ<06Icthb4kv}h-Mo2w=GkVy{>QG_e16i+?LM*|a5o}a)y>L>VM5$-JSEx1mDYAWE zsv?r1p4P3pe8Ke<%H}>be{k}hGxw-%+!0tqSwT`cz2C~FeL|ZRee+97mo2_yMwlyw z+MW+s5z*PYd1Pmx^fxnAS#>7!vTFok7kpDj=50Hb>h0RA*g#y00R zr`5!%+sUp!o}c()RyS$UT;LJV3DKdJMWiR(zg`QC_i~;8_+X@I%8j=WI~2iG7pnJo zi)7m#D^&f?W#Jdrk-xyQzU?~6EcXK&q$ zRE0US&}8>{cDJ7pp|+!`E+rg}p@R($8;)B1-l=PZfz;!7qKBnCJ`@J}W;^>7Ji7_R z!$WO9l&cin6FKrK`W!jVVskz0h_uJs!a#*$rXpkyR;2 literal 9941 zcmeHtXIN8NyC?~fGE_kYr9_4{Dov3NMnNfppv(-tt3-N4q!WS)3?fBQKw3rtaR%uS z5CTLIQ4xZov`|9tHS{FpuFWvtH}}uE&$-XJ<;U6C*?aBvmi4yvuCjOBWpg9GACCOM z#>U2HVtn3`jg6fI{%l{<)azbfyUezW7}J7Y@(Ua zA7*m8}&i`3Aq&ZmIQOed^NK zi!0IcWq{knlk2UI#%-P+Zo4tzrH4nR=J7l%oj6ynkWjvl&DM7AAR9Fi2XjU6a!I?K z6@{_C?;0>VufzRxm;15#Y8L_^*OID4`e7&%fW-nqvI?Fj11#{gQ zWJgi;R5~=;_1WJN*HnGN`LpU(>~vx568CC_X78OS!!5xbcp~0bb)2>Mal?RgR^+d^ zvZ=uqM{3wYh=|!((_>K=mZS#5NVY$}JSdvMN=jg*Bz73+_6Xn`L*18bg@!Jx=1%MK^1E1Cq)f zSq#oqF7WiRI7c5oSd-b@4NIZAz9SEDe&#_+9Up3h3a$6}ormW*3u3PQ>&Zz$sbX%2 zt|%9@CS0nqSA7lcbjW*EGjIMMf&{>w%pUM(h{emhIPXcswa;7g;OhDacTZZ+mpqxk zDm8tW@(cG{=lpE4-5pa>i@Q8*U#;#+fL9YsT_dp_5Y@H5BLcf{HH)Q z?!OFvdCpS^GB`ZuS9TS3@r!={w zG<0Cd`#tqrfht=_!FiRa>sY#OoZ7U12V+^GH4%Gk=djK8ryGlxNcp>pe1W$QaiTEHbA`S} zC^H&M?@T+ZWlDZO<`|?mrV=O8NjzJPPYDWJoQ=%7ZdB0QI0=%&P#b>>cg?PiG@VEb za=mv^_nklKE#`kwO2?V*+IAm%gO=lbEqc^n3SHAT_Rzw;GJovg%t{29@?7O$cZIx3lKR8!jLmGJxNH`X!lz57c2R^0^7h< zW>T~5)N_}nczrHN6hFqQDJdp{d7#_>RY^Rd_RrDz?ORl0|Cf)#XE+H5>va6MG~G#^ zKT>)%$%`}3b&sfnoVa)|@+Z@YiwdS?b+j1VRSw-_&+8TV@#s;yxrQoZeP(mBS2-qL zo?6v25Mn^u&+e3M)}Hix+H^*Q;J2I02;!k#ww|O`cbfE63lDt3dTxOzMV*3`B>tW) z@T};N1<7ldwKRvWb54yN@?6ksHIM1Ou~ZkhQu9K)Sm8yHx=zx!+uWO!YX}LAI`p%} zgfx+>pNA5%7tM}ij^$^AmalgDg*7oc9Sg7-yWv2bgnxcnvXYWr94w z|BMf0{=xlA$?Q2I=(1E8zK?v8K3zu}?ccgJ%AL>MgMv%?lTs9Pwp)X58$B9mF}7Bo zoka3lMC~t)MCej+@gB#l@oa7L3x;-s4<0A14EQw&D>a<%ixl32*FEe)4x6jSn;t{& z4N$5TObr;_h}6Uvd|TtIwo75=VY!4pKIL~ZE-bF#53tDP!k?n>-%=U@K#fmL;g;C?02aM;UuA=Rws4!05wsMg|TpE(gK3tH9 zt^CBX7?lgv6B**0gPqG$eio`tA+p_?JlVXER^xe(o=y`V<$H|YqJ~@emIuSv6LFjp zOC#F@O^n<1=m?GB^(Qv+4si#0U;;W+eyeVMJVr6#7m!lwr65t0VPw>JUE5Z{#_jhy zmK$L32tw$7z{Qgj@5g5Rnl94k0BjDh5#EY{goAL&Yox-EtP)iB)6aALAmHXCU;?Vt zpSU0!`;ut!55KI5?0_Hx?!d4Q*2_z&z}J&-vErCbniT~MLyKI&HLw!o(f|^^ZXrk2 zK=eE2;%g6cU3q9`+|CTF3`D>B8uSB)=H(($P-t>^-t!4v1`WkBkJ< z*(1yB_H9(V;%#WnlY5~?^ctxhFaQ~a-WAj^6!SGTerpx@G5a0t^E{aUh6`zwDB zrSWRay3b-Md)H7W_W&@O`ZiyW%42h83{Q`J`U^|$8f%1iJ`2Z(weRwf*4}w?d=Nf& zANTWFk}qs%IZPW}i?3#@yLg-moQH~bTnb%AcsJdAAdEo@ac_=BGHP?g3X`-3xqI%9 zipo>ZGZq+r6Oxd-4*InfOo?WqSF#RUAVF2WB1{UWnv_b<)3=Y?|;iH~z*jrIwb9&9;h zPKq#=rZ%_GXx*(pcSO#N2g<++Bbpz$YXssD{mOrr5m|6ZivRTO7(%yEw4SSuI00ri zF{4Kzs|YU5#9)nREbIP=;I)C2QDs)=lHvKkTk=56g$uA4J}0>c!YPZwc z$ST{`G8B-PbvcoEHw#Bk)+aUJTR2D?o!;`~?+Z~oSkycE2J#jpmoX>t4EUD?=AWhFShWS zp*uTxk=`i=!^B;)Ko&;9NI)8$hUbjgkG9f&y%R{QCASmWT{0a7_5a;^!7R_o6~|>T z65sK{hG{u0g={v`O-z*ulR6mR!NJzA_a8ceIsq zaEdkj{_q8ZBe3p_@#W`0i;1sXJwuY4ePcdLXW>da&to6pI=O1-VN^b){8v72dm{E3 zkw9cjE3BqQz!rV!lAB&%JX2EdD(GL)v)2?@Y!V5)F|@9=6xTUP;-guUdD{)qh;Tu! z;Kot|3gwASVm#7}6v0p%eNC@gs^rqWE<{HQnW;0CiZ@cxx1gk#`fyK*3AXR>*Z^9 zK!;eRxsE~pn${{YFz%q$B%cl>7?Tj;Q*Zg_?MS`xO0YwXG)Td z6gz7HptIl>NWpVYBGIP4Qqg)M_@! z@$fSk*H}U$mH`bezc(j6SrhfNPSW1Wv# zFJ?}n*Glc0Zjqz^rkzA!boCd!|lzy^v4ma%URII5#N;*9U{(Vh6$1jTZ4)G# zLTCd**TJ=2Ggq9^&+P8`(RAFh~s! zWH?EuF%}oGtntH>{VK5dmH*&A-yT=B{J5e6k);xud->@~mQ8zR4o)Dd@ybiH5Oame zOBSS9($G)8ya+dFq`i=*8c_WX9G7;19~%G1ozhAEuB-=E$lN^-0ikyxzA$ZEB06R6 zWtwdn=#q|GV$q*wm=3;YG>@P8`G-gi6)^DWfc9*Q@E&KVl0}PbIg!CVrf)zf6f+X^ zum6hLa@xSoXD>3I$Ifa34%rZgKe0+;!kDphs`>n|f>t1uRi>pp-JQ;IyC?xSe+pWD zg;h1UYOe6&n-ujiL{S3(e*fcFl%ypEs$kkIg1W~ zt>)sbomuN&{q?hF$xnvN8#lFB<3TB43kIg+kJOCK8zi^&`hQMHOp2cjNA~t#nbOWZ z1k7+^BkGnhfxe}pG+VdfMn=|1fDvb?pR--cWBnF$w2(%dMAuWzGH4c?9G$kYO zEVmT{$f{7=BYqdN!xx!1ZhvbtQGhQ*(Meh|l9TT~1-;E3A!XvYYd9g%gR{RauoC0# zp)MXqJ%izo8?rrC(w?g03gTU|2I3QD*E?>$i3esF-sKzNG*zhDWRYwI^~b<5Kx2Z7 z8=z$a*I_Ua&wIhAn!XpJU*M&Iigl2R_lyb!OMe1xJq+}M?!CTXTMMZ))HKL=Jv#cZ z*kCB3GN^9BDOG|G7CWko;lEe`X1C)t_8j3ep(xh<-q<%!H*i0HO2ZlWOKNiF_mX8= z6L-q9ECR6))#N{xmn%~TPB!2_x_c)Mzk{0-iuf}RaeJ()I$G6vDl6BtrW~FyEO@TE z2x~D96oW~>GDM|Zubj77Q-goxw zgS3XPWBq00VOHO(3!Z5xgns1>?p1&ZtDLW#i<-wLtU^nstmX%j2=Taqm<#20gY5R^ zbJ(?V?E1vR-hvOHx-Ip>+rV~x^ejCDtWYT7f)1Z!HNe0%S!kEiH||x$ zvk!%7OETv@2BX^egVoy;oveqjx~lX8Pa%v-<6d)=pN7#J=icz&PJ4t;oNiQ^!S6wQ zQQeD@6MX%u;lgzj+rb!k;yd@zA6;{lLNO7wDlnt39$wNCgK&2c;Zx4ETh!cc-!u#J za!w+roXcv_VayqZh123zy^ki(CDL!de<8QXH*?>JjVVYVN2%r(ez7m^6{H{WMbdmRIm?#Z~TZePNZ*@&f!8>eU1~$XOTGvOl!xxqKLX zYZC|rg1ZrdoZgZxv~UvSA$-_K@sEEU^NKCc6&0;5x|l;;9%Sgd`vX<&U$6PT5Y z8!6k13=jBA83iuRVrLu$F{dMZTo43dOwyVXmuAKwLFH}V0(z(MQcmp1EePMwUer~V ziBU~?XiPE>rM%_x>LfEdEh5KQw1)Q0JIBE~4I>Wuuzb3p1{Qf+$m=eT7MG|8$XhX` zf8s|%8oKebZ}DG#q#EDxvt7NLVuRpy^s!szF|Djigtl zXEZIq-hh!2(Do3f3@Y%g-}Q`JDUS>dEi`6oHhp0tD3|ULdwT@$r3CDYkm}|Fqwdz0 z!f;{iayGm{&l#>R2q~3N(IHaR!Z0tVKSYo99H*v+uCLAtV(w2zW9M&jJ!&{qTa8QU z4~l`F#@HL?$-NSs@}Wy&QwLjId$sNr4{n&t@;~#cT8LCNC23wxz$#T-*)q;c8lBQm z6;$ac{$kO<(|%v)2Yh0MWFR!lp33Vbms6`Gw`E=6hVpexhkEVyjfNI7cHugdaHoz= zXCCGfyAX=L_Byy0U5J-qV=16m9z`*pifF) zb|h;Dsg%;^*sVNpd_lP(Y*zsnv}IjdW~*jH`Rqnh(}=(mzRdc&a)7f)wVCIJzwJmp zI@Lm?-U00_c?YCPX}mkk#C8PsZ3g){^yc3*!p1U5T!L3fhd3hQlrKwDqo zSM+s%Ia#}%0x>T6Lay{wUhH7#Cl;C3__nX$!LGrBrN9NlP-1VkogOlHwzieA6t?ON zVHby}Hm#I~NR)nW<85`v`v^5?t7Vbz;|pJnc~5}PsbfX1q-HVH@kUuoKmeAHGaNFH?TkDw+dZEprhc zuetTqS`1|)C)@F2taEUV+8dJke%Vsx+BprOwyfM3y#9Wbj_7I}>H!V)@0E$R8HBl7 z5&9|U;|Of1;%4A#?e_ZU)82c!;Kj_apIz_#w9bTxVVWg)n-g7t3qqY$uggrwMJa)) zQDfT>agg3OE*Uc*kR0Iv((~CmJ5=ZL%(o}4%{FsU+@dFcVSZl0A|qI%Ket*Wb$v1_ zd>kymyZZag{O<0Bw7yEXo*WyaI8OZ^d2BG;aR`N0bcp2bija*w=st&B#wbX zs7fOjjpzfq;i^kUM954)N{0k!=gZxXQ?Vw$If9vm%3J%vT%FssUv~JRh+I4sCO**% z%}Ka?zZX*+H0A;N{{HLB{p=M_?dC!3OIL1PIC*7kTciYn^abb{dEw%<{w4h1g{Xn7 zRQZyX5?$wn3m|&yeI*KL-yy#Ys9d8Zp(oy-WrKKd_vFVrX1q{mJ7dF$PuKHJ=B^jst zG}4W3&5#XiwR{#Z=&LD_K3t;5Xar7kX0NeEL7PJmtF;IuY+n5lvZ0#*Z>0&S-x}6i zjEbrkic}XuKZ{cT1107GR@s_iu9d;?_bM`=<@pn3<~2e*FRuXEX6&Yfi=#O*g4Km! z1|pDJFGqTffiWI|R(A$=-<=97xG3c#-p=Z|&}JK9{jxTw-7S#}5XTq3%9pWK?!8-o zlw)=1Y+YW#lYJdZ(;nAhm8RPgu~# zmC%blSFzpEL};mclIx9N5-O21ZBU82i2U^hBO zs?3J-eKaa@@|0Y!+HLpMwh@9~ZtZ7vlgg6oLMI&UDS6H#)4}!rNq&yml~!HvNbm+N z3Fc*bpnCfPtGuT6gB7c4+2@xn>?^b95sE1`(|vt%Orgnm)OJXj19rV_z$@*YX;^I3 zK3%aPR_^+|%!!6<=U3~868xky8r%Mq>FYif^<>ZvYG&oElry$W@?&FV&=DI_XVImb zw`#Y8G+E@<`Myo{=SgVWK!@VPq%>RR{O5qWhtuY# z9*(sl>w8{ZqT9W9o}RBd7!rpby-xb0v66%1(9&2uYt29Fx;#o`$kBZ^x1QlV(V{c- z{?@JS@5t{<9?q?sC+h17KPoSK3S@1Ydb!xZCRvZWEcDGi?Wf+fSMO&wykydc8dAAm zt#d`Gz;=*$0g`WpmPy z?54!Zd5$BPEjjsNz88|1TIJ>Pp_YUv#+$%#9k^D-$4e$W*(o-r)#1Nn%$Pc?6)uA& z0!Sv2-bZOCEnS^S(?#=bwyE_tk%|!H2C4AUqgBq0QzHHceD&^`FUvM)7Y$?{g_PK2MZ3Y{>GLx-u5ebj)l_dEV8+f9eA2o7<*85^mFSDi1nX^D!?!N=;WV%d t!J)7OWU(WW&Po0EhduxG-!d^Zq){Bd@f7;)gFa6)xnO?2Sl{K&{{Xa5dS(Cs diff --git a/docs/README-unnamed-chunk-31-1.png b/docs/README-unnamed-chunk-31-1.png index 18fa90afbed1ba39fb8eb32c65ac362994c6b5e5..c44d58d9b6e5fa906b0be6b738cf62eb8a337e58 100644 GIT binary patch literal 9602 zcmeHN3se(l+Rl(D7b_I8q9DYI%HN8Hwrf#tvECwGG}gA_m3RZBC_xbtNnpkcqEZA( z5tU0y(Ne2`qTI;^K@nL&L=6x^Mgc(}Bmp6TTxKVM1iGiY>wnJfzkAN_oJr=)JKt^I z?|q*4{oZfRZrVtiZf0YK!C75D%Hvo5()(g0;JF%Q|Rwns1^8YWeJ74Lx4Uk6rle|AtES5kP;@dF(*fm;3`P) z7Zjprf}pSveXK2_JrDtcplu<71Z{Ej9<8U7m7ce&LEpnPWc{{K3}$w!_6K-{o%|7l znTPRO^TjvxG_f!ew31(nmVE+zleT|f(@Hb*Z6PWB{(G7?|8(f!$xb=)%a$J&_Eub& z6N(px_T%~T_$VHEU;3r%cMESE?ZQ0lIjzBAm>xz`%^Y_Bz16f{qwpR|1P*4IDULoBM&45O z#>5EQ{@f5V^=911a8`d?3!mLy+0W@-jcr;tQ`%hd%C7%)cioOh)o1}H_|eWh4i2)B z^H%OD%{wj`T;`k*Cs~D!u8A^d=F39i{%buB*d~9U_^Ot!#I3x3F?{Q>(tVcA`nyyI zAllM-(`*16EnW)OgMEFgvBfuULq#4x<_m6KZ^w#G^g%94_AMBh*7QuAe|P1I`EC+= zcls3fC{0HTYoX=M_pzO4`gSgZGh!TJlc^6an=e6j0iRW93tHW5Ke%Z2|B&`~MF5~; zq(Z}DXx=*ut=5S?1>z|lP2gd7^;-B6Yt$|Sa?r^?2I`~IZo6woX zBfGk2cF3u=72207S;tHAl|RooJy7{8zdq4qyvhve8<}k__!$}e^=X&+MQESy73KqE zaG-3KROpNB`M}70&rfp`)34*)y%gM~lF<4LZt?3SKIUYerEciy5VuE)oBMl(1sH{= zJ;G@<>&-iV`_gW03#LtnKey6W`3!KY^FN^-@}k|<&E7ig;QE%Qn*JT_ziaGYqqFN% z+%>)R7Q?Laeg@s>ylzs0aLnNYh`7%J#D{%GqYS&b10=n<7JoKPA-wWH!U6x3qu7>L zX%(WI(VZW|Nx4cHeN;}iC}XEY8Z6-&6knt-QMG<&!6J9)HD1-kOxk&^t>n%K;fp2+@Y*gp%u#`4S5l(2QzB-U$@6|J+|YRkA;zqz54$6rgVQ* z{A(XhKt1J+!bp(@oz$8BUO+UgFJplJgaPjS93`1IEA&eG7TY$w{zh=)bbKXk%(g@K z0hHG%5LVaNZ$LbYSTgMBg>%xu(r0>cmtF;bVEQGScr2t~ASp{$EgA{CyKA2oHMKaE zhTKT?QWqb%Na-f?wD9&ZRR~q74`V4NdVM8Alns#A;kraI2gueHsh#4!gnoqPZFoxld6%a8FLl2l zdlndN)Qx~|qEHEY8dtLHjxvxhcxBbvVOg8fTlxn0@t$@^)fqk)#yM<0z+!#BC04Ag zdW@T^<6m_6s=@2(u}|vx(UAoar^&)-rupvf0!(HegvZ^igK1IQbh4ed%d&! z6aYsHcE~DIqYQxik|!Z0&6Y2sag*F$G~-!Au6nef9?&j~MsY$$U+gK8=hy3@uyls> zD;ibY+a}vX;~J#7mWDV%U_Yt#)Dqvwnu1A!R;&f2BM3`rM-Z%JPEXVE#nKthqG)>~ zop)aV4<^O2urxAblAETw&j@0X_AVT=c^s2NcY5O)A6++Qqh^Z5dV2XIf8~eOs7NdW zB`$*xSzPV=Tqym7IUYN>%#tQYnJyn2nT=F2k`x49Si@VTv-W2YL$AUKZke$0X;V}q zl*PmqHelZ%{mXF&JeddrI@V7n0>&W4VZk_a8kit-ba6ECk?%?a6A&^scQsZXtH7n{ z)ujlLz319q8SB_SZ@hLKD5G|HIeUgD-_nK;O?VJWiDKcF%DioU-% zs9uz#DN3v}FJ&Y1V4;q*VsY+=!bf^ZEb~QvGo*E&jKx10o`g4F-bw=3T1oqLqHm2n z5!H9#)h5|ujbQy4iKLh+djFB)-vtW$|Hx$i1Ho0vc3wMIquPzs^l}VvT*nKD9~RYV zr1Gy^Lgjv3140(SC*5vCq7CO%{Yp~fxk@oO0o ztWZb?;$8G`>G~32**QtV$t3xC##}uOZ3igKh2_i|YYTl;dKIdUsb%aQcUNe1F|bTZ zj!j8BsWMXKT-HoE5v;lK*ECUkwT zVbZ;MT_klzb+Fo?K-1zb&^&O`y+Up)92_n}iaB6kVN!-#iw>a4j7yoLD@}koxEPP>zTvx^tHEu;bCK z`-qLglY-42_d|c7Wya`Sr8OFK?r-nn2^)P4cKapIb}XeZ7A$`Xun6|k*!BA|Z|l5Y zvXSPTp1&rXvjJ71%@7f3=#u)};0#l(Qu7eO=|g#<@Kc1ysaJjRO4>o67cowq2rO7SJDw5W~3;xh^^VV*bNQ} zZWy+0`i_@B4T1(KNDgj1U$HARsjzLOk>+l-Bdp?sivhX2%4KAOQ?q={cDo00eWzYq zN-_KStFL2A64RYMjO?Mlhy!8sT=K1cA@Q&YyDW0dJJyBYePcq$;1@${gv{<4)e^vxl8$lNER=u91{93P)4_iWhHUi>3lbQ&J$ z%R0VS-1YSrd3co2U2cqxDr2B}y`wr~T7;G}O7@J8Te0F__(;v_WuSQ zzU$pWxdkJB4WWv+Zy^N-iCz3&wC~ zWz3W%O9vKG+|lJy@YZqs2J+e~J3X8Ms-wX+uW_ep0%boWod;J@9SqVMqyAF1V;-5r zAQOB;T(=i88z+PAtdTu8;;0OBJ~*7r91hXM5VQf6VNlY{AO9vvVgTIJ_-K1%-)OR= zdC=N}gYk5PVLfQ8sSeZs4slGLMJf#dHCftqoq+?yPVgZ^pc9*Dpptwn=LrEY*1tYm zpQc@4=?tW335jVH^ftWp1R`pN6%|$OlqiXOCAxxY(@%FPL{yiK=3K)^sFz}aYJE>0 z+jx>}@NN$iZWrPgoBCwi6b>Yhx+EG+AwYHl&3MVrGiv=KhE{zjzz z*@AS#po&gbuaXRI?Iui+z2eZ7Ni9vDn|G}DPkrG3H+2Q_@E}9A`=yM)#mhY=lhLlp zft~@&HQJ4!s$#=N&<*}%q)$M{kn=IrM4$Ea=R^wKJp20P+MKEZjbHHsj8kdF*;#H> z-6GFGRAYNZ9W9fYv)z7DER)f%O98c}81PU{%beud4O>cq+I+TweFe`PZVew>@I2fQ z6t$U;g@W7(hHEcMocb0qLQ&aRTsRq}v!sPFo5NFs6|*_F?@h2mGny=>!C)o!s`v&= zHt5LLIGu$6yI$&EptxBIU`4s~FHV>_#29k|U!0gZXKAY@(C?$!qB|j>SYnP7bmll~ z%dC}EJnmzQd3)jP`T=V~EZ&>>xwP@2(HXex@yb2Ft3h5oZQcn= z^^|uTt8c%zTz4e=@uNt2WvZ{_@DnlHok+9`tIy>Gt60?fB6hB*f#5m%Rne>LWZH-{ zWZ*;wUNbEM%xJH4qb?+={NE3`;J3;)^I+I&G(Sq?d=B4h;LeoRypc4zSaz$Aae7S@ z;U>##7i1o@XPGOb59g~N1xIF5JVtv+-2U|>VY(;l%G0v*5cMo{+sW#(Mivz>Z`#0% z%>q}zo~*1V3W6OVngs-G^ZGRabTgxC>L}o1lJU9P`h(G{0q;+niBIA#Nr0 zCCVcMCqU23ts{Nhi`)oR4>>7?+pMrrs5_z=pr#^A2r~% zXukzmUK<7=N|oq)@Dgb8`sb?KG1efX+(tV8+UwLC#>{OZ)e=6jsW3}EonEd8A==lQ zA0Amz&ran=Xnrm@N#P95hlj|A)Vu1<7m?<{g88tMkH*ri=TAC}IK_^aRWDxmb;g$6k1e}i z6Pp5%HMD}(XOJ~=nzKF_ZP+8C9|0*!k{-3n)|kSMuFowGSMRcB znE)blbb=PWdGX@qfbX4K!)yCG`FBW*0YdITtuj9u>Ka+n7Sg`>}2j)?&W8B@68NKl#(`-y49BWblWt# zeOI5KE&iagK{=&-fZg4U~hroeW|*=hcOSIA-V9J0khD z<9fOgWgGY^7GRyA&Z|zMovTDj4vw_oIG@5noN4O5%SGGEU)-VzO?!f7JLXh z39uPaZsDUlhLJSn)bPtMe0=QBE>ru~D;s(UgSl;IX|X$9JW-nmha7G_$7V2F%kl_O ziQUb~C#kq{SxX!(9h{G2y5Z6~PeTMewY*62GkrKECq9}(Pk7l;xhv{SQm3Si%Dj1q zB(asg+L!6Vo$5Jfi7z+=rvg;wcfAr`+zt9q~Z{ z*m078FSsSE$R$P~6vuKjP~`h~c9oF1=eK~D%Oa$YAQxgm9L>WK7W_ATf};c%?DMq@ UcX!LweoN)GcH^4dRl8686#yVhIsgCw literal 8890 zcmeHN2~<PC`&`+q-(#TU~d(yPjvAoRjmPz5l)U zzyI(1|Gkq-yLNgS87wk@!C*$)wtl+@2Gate4@^%Bk~9mat)Sb>UEB9=hVC#J0tUmu z5C|9ox+cPK4Cq<)q&jd29CRikAYmeO@^Mg61av}TnCgw{NW>*VCj-ae;~2_B1S}DU zNQ9&+Sz;oCpU77-a4-fA!B7csDp?{!eej`|3_f&KGL(FVQdvwS*5>B&6A}DG9G?N5 ziF^hFdaWv=I+T2+64IsQC#s4=&rm(JtmK;2D`*`0p<55&VK9p{)d$!Xe(nYgwgk5A z+YNgO=>sjNV!nTMGrrBid0*(J{M+%yMV|`|*B$n|RTLL?)Y4w5MBU6yS{{VC#H!vS zKlERs>BNfhUCST#1|0BAwSygA!;A%BQ*Lk__%gpG&Yy1oQWM~f#-$Z2WZq~}R7_W~ zJ;jhi;Y>H#M@d?P4#U<41LVbwM&qVz!Ps|%;Wj+|CeT5cLtGU&=Iby$-gE!6vWSC= z)?dSbD=Jwx`K+vVoFE_owst3gwT_^NI>kqNjOQ3FDR&SC=eeS^KuaFD-jBqLlNB2o z-gwEXUhfx7)uKgO>jq~=Uq1{AQ2VgUEZD9i2V zJ?jfbN3(u5hD0qUgfOI$Y-(cQu2Z}$y6nqa6jN|e7fJrLqq7akCxzH~K?;AQQ1xz6 zBCDVmh{4uAVw1$~8$kS@Eha?+;rYWZH)!!JYtN;A+PgM@?V=ESlD{?T{yUKYq-(<* zfNCQ@q)(vpP&peRRq+)5d8KFqAsc>?9WE$-q#A=2psXI#HjLFyZ^@+GrBVv{X3is~ zp%F3gcn^40FdD4(1RH;6oMpoP|3{VBsc>t>m{f7@o_IoK5#}U-_3#*XmJ4NvR~N`i z1S(rB&?XZ6=Su>-J6dSgR6A1ZDaeHElW>oa)QkS@DDt|KI`qjvs5WK?1cm8-v#j<$ z6P`*P#tTfjn2_C)(8emXR++!fOfr%z`E|8GH6`?9if4JkDL=TaFk`nDI1nWClycKu zrwn$0?mxy;(r}WImKULtz==xCXE<>zT{Wq_dW0h*;vok?g*iRC8OC1fcYfDyu>N=Ro&7b0s|j`QkAlViN7fwT zm%zO5xNiwu062{;^E?0jQ83FKokab@)cNUyRl>~Qc@btCD46z%?l>_bcW_$qg|KDU zX3$3YHQTGVt?aOGH5zo!il?CcBUf^<^R|Q9Wvf@}jh6+CwJPN0^q?PV&4Zun0AW98|<)--~QR^>Dz#_8> z@YI#3>X*y>wCJl0Xu3b?nOvb5mA(9uXq~te>C7du6~!a}UDJ)HiN2qo-=MBHH6b5Q z89KDbL}VnrwpgyI_9mBB+nDkCTw{fvz8h*wo0Rgmwz#NKeCszc;_0cQQ`et*CZXqr zk~9rUNnROr&%(1G%B~$*S`+Mx@R8u*zOEXbL2a>j`Dgjr zN8we}-Q&n5n+blwoQ^OTVGKfR)isS$l7cvcDkG)Ko(N61+v~RK_#XY6hsjn`YYV`Va8*-S zJ5abr(|~Dy9M9qcS8LCC-7Vm!!vEgZ=xXJ5GihQQWGpW-240U)&HJtIQ7KdtkL@1g zu?^e%S%Xl{bt*%JoHusDXYNT9G=HghdW7IJB=zN+<=3mdn5gyJczaNdJE_t5PmQ$L zBVj~EThoC4{yw!o=RrxRV_pQ5eN{z6!EL4}-k1>WMHsp*hGMfuVnhflE7GwQPHFtD zE7o$nW@@#_h_g{N3t{FTgknvn(kDJ+MImG1o~w&(*~rvbNz3~lTvJ@};1H1}NP||rhCd@NxXYguelujXwuFGr zX`VIR=(WY=+RbDQ&hdSsM*)Kh{0r~acaEzy$d-@do20Z&2ch)mag_?+T@BoA99e<) z2Z8*T2*vc;pX%NlGE5$QKtUu|O^HKp|K5oB#A$K=TZZpy;)+txgUrmS;jtX? zn08LM+Os(%wtCshhVpfdh)MD==3D3v)P=+>N6IBpP*SX+2497bVHyjQfFdC@iMxhK zdPRSijA$J69k=ta%s_U8AlgKIuOk?k%}O-Q<;8sn`5Znt*{m}-8!k^fJ$et&u6k~i z=(Mowz=!Veqt$uJ>)A-`O>i6mxEH3YG_ZP^{e$8AY~qhW$@fL6vX|+2acp@IS%I~x zr&D>&9U|rOX(cz0+;Vg}ld&*#@WG81PPLI>a*#hg(KIwNK2diDM2$(`DwX{cJgYA8 zgD~(CXLmj@FEpJV@--(YgJy^Io=X_{HDI{lr{~y!N^Iy{>hkL^$*()9-pZPSudcr zc~W%QgWkJBcy-~^=mmyv=ro4VSGo9Y@0fPktBw2VXXDn6;otV9bX9#MsFmj4!7x_{ zcg)l&q)AEQIjo24fRKmt>a+d)kj8&O|4l>hu8q#!1nMj>yh52JHJY^t9?O$&uT_jm@|0+w;h1Qod z_-cj+r#`8TG=KJ_uxEv1d_pqsLw~%?-Qks6Lr zDs#mbxx?f7x6nH4FcTt$!RIUAv%*KJKSouaf;hVdom}_Bb&9KxUf+K>JF6l=36b~i zoG<G+;eHvcII^}&3bfG>%IuP8LE=@ijFk&vkp!TnT6Pu{yalN(2DDAl&%$2 z%4?<+KE^9!aKYeQXLQ0$)Bl9MKQ_u=!NUK*-k((e>uCKW-&XIhXjlh} z`mo8O@t)=h`EPexW{D`9_M7b%4V6GX|95GQ>j5=z?vWf#N-;&|&)jWP3nopI-e8MQ z2+6kiJvI4geTe8&!CGRxWzq8){wXpqc^}|EPIENP^6OU@7!z#kBW6N5GikSbRz#PAn2AQlP~IS{IFT$EWCHw zy?}Bkd=k(aZ7N%G(ZIFQ`hL^PIZ?;l{K?)247Sv{1P>+)_ALZ$gXKZTfxusrvDGGi zX3QzZLzv6yijb4Kvky;q< zXnW~)sY$DB;*u06ndt33(tza1AJ>q{MXzIX1%tJ9=+p(&WKu`%?#WA;6=Y9w#aaYg z`s8@|$$6n)C7egxJ(jCK-Y%)4>L3M*7=T7qBBvMXvYbQ#n?M(BpYPJD?Wy!~#Zxyf zIZb{?&>R~z`flfn$_x+lw{=jFE!sY>m);$^IJu5=#2w&9DoJ3hW7fSI)FL=zsc=v7 z{N6@EGCWfuiG*8Qr5F>4X_Fp+rD^C_75Y2Q7F)E+Ib#TaaropVlr~jIbkCVDu8%IPqOlo$*@d&I6QCTTnX`F2y)al9hawA0SRCVyXl|p5zqH3Jl7`wH0z7zEqa`eN+66Av+W{wW= zkf7fkV_q@P>ME=u7GLyFa>#?!_4HR1qH_aDiWVsaxB^~h@yVn?;orWdL`o*E`*%heh(c@<}=rzPsyoyh|?unt?<9$1z z%eAG^uxobG^P133@L*+RadRg{#My*S>a$O7Vflm7SiY6NIj-|EOEEIl6h@3gq&X># zXoB(HUTT$-9>7=6vWHbARaSQOrey(-8Gwf7gErjP_wZ+mzr2!DfKPBMjNww~eW{M2 zvA0G0#5?*_1Amo}T(ayMC0DQLM(4dfhA&O4LM#{$-eg(W483UKR74&Lw|;j(pRiEW zfzNm`Tq>JO=wWxiB4dJ)%i4)Vr*|nYxH&34NhT%OL-LxwPW8|Zd=W;D4iuHgbZ*-d z>0zH|mjWO3xL-J5@+h~58Md%O-0Ch2yUw#kcP>H_QpVH5;Ox^i$>foVv-9$efRu+s z*SF6jwb&Mpx@E)9vH_aY$pdDgqMkGs+;^g&_U0GnFR8&-HxY5h1i2?&GyXm}>Bbbh zj>7ZNVxoqLV-M2q;v!@vI@aXAkoIzXir7RvW#lePzDJsE=DJalnP`F!X3HEZQ%-OY zlN$&1_oA#NkO*&c(au$XQO5Tu^>U}WN9TO*X9;&)ZYdc*u2pewEUbl9T@kt8A^8MX zq4Z^@Zx`xn?QEW_Kn@*}hV%!r7ln$lGi)-j^Wconv(THeVKCsD>c0TOJ;MOj5R>OC zGGUL9784$haTCgK@5{@^mntRgU^Wlr4OP#q7Axhn3HyZMk-iHC0zR&yU0?hlsPkZkKuf&(Qn;$NJ;h$bDC?gTUF5u~t) z8HYPyCK;7p)s&0QCJ{YYu^=zLKEuQhU^8XJT7yQR!H?nNnG&|GF2D*7>Z zy1QR9C$_>yJJO&>Ix4sA=wRK}ND;N@5#~h&z)Wz;d|Y$Do)x1^G24l~Ix_h3c&2=! zFH~OB+*w%)L^>T5dD8V7u2JGZJqx`n{}*okC~wKW8njt-g6cmg+cxj~HgBWf+5Z8| CE{@Ou diff --git a/docs/README-unnamed-chunk-32-1.png b/docs/README-unnamed-chunk-32-1.png index 27369802ccd97554d5a34666a5eb7a3d32f05497..dc9186a0b15d0ea968c5579eec9bad7e6a788e35 100644 GIT binary patch literal 14910 zcmeIZbyQSe|28}f4T2KVp#mZdB}g|YC`gDjQX)f#bmNd;rKO}nx{(+f1{4vHZibfb zZkXo`>V4nqS-BLD325AE*cv2$3Q{ zdL$5#(CgCB>k`*_kx(cIT?G9T3H(AMfjb-ul!OBTfMgtrQfV zsn}>4)R>Xcn91 z5`~%!Qk=}NoXiNC%*dE*2%2mF?hR{^B5RS*H8>C=*WmCqWaJtWxFgq)s5KOP4TW4o zp-^xr3LXj6j_wdp7tk|weJIxn4q9LZ;oCpeaRh;=lF%2%GiU!85a;k59la^5akVVJQp>wtotf`B2Ls?uKb-S5l9)@LS>(&6&nC${)6{NJ!ADtu|uHH%xecLYS|j-VEB~3hmig-uHiG=5pj^w$x+R z`_ny45F2E+J28y`$|DA23f=hsum8`C!E+?Inw}A1QAFUh|2~XAEeyA$(TryLtI@l) zv3;}c&++qP&J{a&H&ijLt~PFXQ{(#Vu2Z(Q%=^pDkx-@Mz%Rz>J%YKrKME>)_~$-v zW#W+{vOeY8+ZsvrWKEhTiN1bR@ro|_2SrW&*Lnj1ya)t{zwExf=IBLY>vgVz5h(b} z@eQv4GoyDo85A@KvOHcF50Vi|F~FMo)(--4%>RkIls0-Na);;L zSM>Gcv;tM~%nb`C5B5pYDH|&?_bsfEgK)0~b5=~ouTA2ashImkrL-O?Sa#-1j`|nr z@^NmW2t=P_0MW;89}73Epj0O7Nb4iTuS`4#bjJQ~Dhz2DU>VDJw9HAD13(kS<;U)f zZ^Q@58?6Yp;&ecpABllwdr1Ty7zhbPt-qd2ILIzIw!sQ&GFemQC-uFb=_$>mvFmWHj$VVS@ClfLnR zw)wzesH#>uu=4HiAyB7z{DA254Imnb z&Xk(ioIP@Pb#r~+a^v=&ajP6^!8a}O#VsD<_pFxb{xfi#w?)=xq#4uh@w_7f9#zYz zn4^DG-3Ow$CGb>HNxeh1u3hG`mT%ccj2uxda>Z|ae zhjht#eCbv+}!&gbZ8}iKkp;)T4#{LIyc)q+DW3*~DXMT5vN1Mg!-~aRl>&h2- zEroY=Ks>v(|7jHB1{9PRQ6{a+?nD8@D7tGM88@IXJ9QU~IrL=gj(CC(4#Ptl{B(g< zoYjsm%PB;PGMA_Spwj}VEC_h!gNqTU3q3%$s1UqB$DfHffX0B) z_%FYp>Y!>eu=ZnMFcQ&!Jo)z@cf>o8_!t}atRqtwHUmH|B*1dnb&>g9wVUYOcc$po zq&clz<7;a2E7X{k3;o@+-F4h(OxgMVtXI^_TV#rB|1||VS?=?mwx_7^%H}MaB(9xJ z_rb6^{jHOu>ZXHTNo!`(B4!rmf_KH$WM|V1BKu*{({p6g-ag}^?k&$z4&Y`7E)}KI ztCP|7WsI_Sq!b#65hVWQ`R1`7>h)%I*abIar?)BR%6BGyOSo#sU(2fV_{vN&1aU_J zKZMuT>OI1Bt3gMvN$VCBGuy8h?i^uqACi)xN-EGW z)oiS$OHsp{lxkqsb@u(7LU1$c719%QU%SsLO4_B|A`8e@x!11-o^h`#L2Qe%?`)Rd zD`+NN#JXfBdZoi+%_RgESHf5>ergP{_m_La z3_XMsX$yypvc4?S`|P!)Q;6`m>3yaUxzA~DLwEbc_lh3N#UJSLotaQ_VipeZ+^NMp z`Y}P*G9LvvJr(Sd);uH^#LRa01@<1ju$PT%aFdTGOEB5ijZ2x6{49Efa#jkk?+!~6 zx)9R~cY>LpW%QF0+r))LtX|8zRaJvQVXARhS6AC-7O2lYE7g7rn=>e3-;mx*>Als= z2y9fU=L%($`1t_iIi$yXtLin3PbZI9ub``))puV7swZ~#v~p_MZ`r;|x)SqQU-g9w z<5{;|Z*R8SOdK6zb4KaxiW3~FDkB*)A zn&HySan}Rqs^ByUmt{$DnpZoLw0iT3v%VMpTD_NYQ3yt{v;5Jm$*WNxivADP43|Tr zUhuYW>U`^9Zto8zEQQ z)>t{Mvfv@vqv*|Dx31BQk#6=4zZ=Gn7Me385*2nh7W764B1%h)mGR)wre@UzAB;131V%L z{e~3{0f>YiA`L7b<`mmkv)7N``Hf5zCuLX@e@b*Kmd{da)6V;zr+JIkS`iNza(WZe zi|Bk(+~qvj17_UymcS1vR4{E!dw<+*HqipJ{2#>4?jSr=VeLm|+S{71yUS6tU(^Tt zQ`BGD3B%GLJ*FK^-QplD7hGB>C(jREWlzdM=dK_PqkG*?^Xoo0i#F#74*!TIIk|MI z{;riq7^Xb>6#I*U#_N1MVCF_6xh2uhih6YO|In1P1(s2t*4=OETjR# zPE(X1jE^Iv8x3gFPbYL>n-I*z?6d99=o-?ndM_QfM{R0gYCc9%k%+KM9bA83RT9&Z zOXA3apYw@W4oCcq*78xm87fHA?!fY+tooa#L{ipHD@~G|<8O)1&Y_z3#S4KtCxOoj z&$7j!em|~hrcZk9ebvw%(eZE56Wx)YGzI4M3LH_^3!5p(&hHVuOW!}>Og~A=lZN%@ z=GPA*zhp+_gSRsbPsqd{U##}F`!rBPk1x5~-ukDU6f=7X86O_U!kejPXmOcfcFl(? zMell-)EeyMaBsPcow-d>w@4-HzUCAJK$8Hj8N@VtTZnC#QiQ=o6ha?;%+KuFFX=Wc zI2)Z2Q=x*H3!NdH?%1gv^&<1Ry18fERQmmAVT<3h)H$m2(rz3AY_X3EXMLlYEiKY- zsbdoRS1#>(yG*y566g?mlDWMqAo}c&s^7_pR!XHZS;`1^LJ3G>19)@H#q4i?)+%KF zM*>q61eSfwpYfb6nXC{THX`#k-f_ZLyFsLE81`^Wb@AUQ{(#OgxFE7Ia3p9gt}boF z^L4S}(KiPWR?t7B#2ScH#9$5gL7khSIj)q^KM$5wb$G8fW)w_#3pM`2NHlub&PU|; zP?pXA0p^Y5LaT43@7@R_Z16Qz(qNTTwNv02+-W4IG6HZ&!oh-rTA=BXM|t8OG?Q@f zLu)NHER5(EWx&Xd-vnSF+N8Iho5PKj`%~lRdgvc6kV=>RVd@3u6CMDrqj86Y|A(L7 z$%g+vqc&llwbf)mj-W?xgE0tD98gGQ@5T)y0B0RASM{OM%7bXlO~iU7Cgzw78q}m} zgEWD}YAC96qK~#HC5ll|=!wyaWGhY^VKhWVKy`izAUkOKF&~&d?7DZd3gwhDS`kI( z!w^abQ+_I#i}*|ZK8PFYx)bjQGtE@>4GG#-uN*d!*QLWs2DbW^ZuHNQpKr_1fP z{}?p~@G`xzeKSAIBf#f_DIM4nwfd6COq~74+u)oGXj(G8hS-7;J zp#~M*hI_v7^x~h6@~%eF$O?^Hn?Je>mfc1VVK^oLA*Q2Nr*m9i2^^8bOob>1j%=mg zl}URzkUy=zh7Q|Mu`uKtbA1seR9C??pJCPKb)0%R{PG4Dg+3h@LgVl+ggmDKZX4S<-@YeDqUK}?%g zkkTV8CVTJDjkdXwTfj^b8z`?(A_^eEUJz37Xm&uO&?Ipyj;Yq<*zZ%pWuwk|fkzBH znqCQHf&X$ut?bW0^wGbfpW}e&gI1i|MfJCtP~mo1CGW6-R|^Lab!HuN?b&82n{*q{ z+2R3cC?}w?3j%t#c$hb`+i*x1!?3cj>#!9nBNOhJ^I^I;^ zWedGYa+|ad21Wp>F0byZ9{6+|m}-sl#O@%?kFom8WB#60T-4^gjE zlG!XxCFeTbgal{!OT&^t9Gy-x*A@CEhN0If8qX{y zZJYpCZ<@GINSJV;Gd@hfMj>!aE&S$AaJ7xe?Fk7p;(OMr_w^fDHLG?CA@)di&8*GA zC5BcUfM6Ty(0cgu4>qaue=RLS*o0`u!h4fE@o3-hRV1lc7oA#*8HrK$b`ng&gNH57 ze+mZ^RZvj9;bh$`R!E&g25M={!a0j%k;^^&8!P;yXf`{@EdWrN9JYPSSbF|Q<5|jP zpSyG%G#bXGIS3nwb~Ici&E%7)df>OT^eOmWvUbb6E5(Y1uxNt<#UFk8FRc~p}4`y z!B1ncKtD(A#%AMb76I?glBE(=$JhG>%MTPtw^_FyII4BRH3?>axWC-_{rmUJudu{^ zt_AZ1-IZ0LL1Ma|n~jey2=FK24sjOeP9pxbvKpUGD?|^W*mRh&1KaLf(KS8 zA(scKJD;kj#z@AGj)Z$mXH7yO$|fB zmHgnn`Aq2j;b%^N8MRLaqOn$PLR0A;mspEyn_Q4^+~B@`y^zO_KFhCGMv?o;<;mv7AyncDy5j{ zSE*s1aT9N!4cku2hP zw_|m>)U{~K%KN0fg_7;F7%=C+$9`G1+gl8AN%VlX-!CcSOfc1*5>}SKKdANRkl_|mu1MIVU4;V%l(QXPAcX* z3&zqT5-v{+H+GHsQ121h|SgJbC$$z#vHnjA z(KQyE(J(W#)x>v2Erm1r%wtE2(xM@mIJx$TLitYhCgk(7K2}TgU2E0y%F>+v)p^1u ztGS`HI>|A~&GM(IgujuNfv%BW{KqhT9Ttw1qo7;jx6xzY0UfwqN4GKP6WujE>m(`SuwcFyfm4FV;fqdH^R~hb z60krjkrty873*V&!2i_QR3p#lDcre@}< zzhGP*sgCavmnO~6cZjlo;Mq$3Cts38)WasvVUTdS3FJY8Nk7j#Ww1XhkvtjGj%6Rb ziZ4;p1=!%DlA^wf9=)l;xD=J$9y>I8ufXwR0ABr&lg9uhYZGqfJAhVn)aKHIq7E~axM%6NWbK+RdyVIxI75FjI|@WsZ# z5p{Ft0A3yd6E@(kFXj@=&Bz-47=$<3D4A-jj-_7xhuL5??az$4t8+YHr>Km)?Uc zmhx@O4L}D!ur6x@VueZ1VcKB8;cH$FTFd?^uI)^{4_u8ITAMw zo!h?j`GE}XXHxSWIZ;=54V^^+&775(rSW#SBU_D_KWz~tZf&d0&9(>MI(tB*6he>XlBo&r{<9#?v0-W|iX9xW=aac_n0-0}f=*L8@ z$QfYLnXXp?oIBHPX&%C#=(@_dHP%8_Q?=I*zWIGP=$Z+_o+i#Me_4SH$ZFY-&Ojkw;EmFg^uNCM{Z?>{!ZWV~AZ3PFo|ycQ#Y@>z z7+QL`!F7f_Fwg4F;+Dg_y2lgK#_vnJtx$Vhvqn(sa>nwT$_)rDmNNNM7XV6Fo}VaC zOzN)fNeBY8Z;vp`g(_C~JxCH!lWG(x$iN(!nHDX8B6b(OD+L_%18NO``2bKySQNGVv=XuH z1PF!K?((3s6yjWh4Jb}GZQp=El%LSP00!IyXaVOxwnOLDs2eY_g09f;#&Z6P*3Ey7 zP`irYj@qo#A3MgsHBEw%sSoa2Xij}CDm~lQe?BZT2q;}>i$&ISTmS;(e9gfB@XYuB zG8_!GyvGr-Gykgy{qO2dN7M#$%;_JkPrO!?JSBja@wXb6#d_@8(DKkSv%4BYx$R|+rb)PNk)c?QcLO^ZEmqAkFQ$8z$-QGv!SDD( z=?el_V5~|0x$`W&qN1_nCl_|UkET2?_pb&hn@)XwPfmkT$AjaX@Ek_yh}WDbd-K#} zT-r&=#G+)?t}FP#x1kKnpBLj=!zvn9887SD{Ta#0PxxjMlnqo!?a0kfcjO2Cxy?qf zbCyD8(?JZQmt+_}+qFL{Xcv4_SYM;vplnwILqwZ-6;o$A35$w;7I<6{#Tsi4+uOiv zTmlHORj~TR*AlMnjwb^>a?*6k2SgO4W8BXUmxoon1ob9x0&RNMiaS*HBI+n})^sM6 zNP3#*mV8R-oTqeOtJ*3#P+sS*sv?mG)xKY1e0-DB=}ug*$2} zH|G#u?8uu`>5S|b`*RnX{W_@bZGL#lq=l_neUEGFnBeu6>TL(iGh8zRv4=vAhaXUM zdbZZ6tJuSsFVPS=J4!A#`xl4mah%ODoMSzdLhFavhGz2EJ}08KTgU8bou3EZ7!;EZ zJqa9i*m%T4m7j%Jb`)%Jm)M<}%GJ! znqQ-OB_)ybnQtqnBK$1hR<#s8JLQ|X_;K?vv_SZLKX-YV)OHCansUKYmB1ig-+yX% zKlevJ&4Gr@K-z`9%rK(~QuH$^GR3noZ;yAhHEC`f(O#Uqs$kTJ=cA1;BtIK0IdR@u zyN@y#xiqevPlz}&IHtesaoRIucfSfLz@x+Ud_?p*kW;N+{4Cd3uNcK+P<M8-62^GdzpfzYTY7cm>T`1HxSV#= zSFi^qKPj;ANV*}5)}68h>TjVIn8j-~ntlmiSMK^FAXL)WH}u8Mtp@!9YPPQ5g3*uE z55*Q+6e2Gl;PzI{$v_8HXUEsl%L}!Qd^5)`m;Gs%2=g00wmKF*oi!uG3O6BN{9I2h zw7E=Snj@h;^I!}Q(4y*O?=K)>-rGa7(_t?SwBt2MDqI6c>jNd?J;$aLO4f?=eFB5Q zxWfu|!S0RpzLJpEit)~=V)d~FRwD?0^2dDD*}TKL2k~G*Zu4BJO@6Sb5{q|e#~B}S zs;@jzw}n;)DdZ&~{R}kdahqj>O&tw{Tqqwmq?&`&4Z>vqiuC^b6WdAE*`1CW9-xl2 z?#)|ZG}|4&9qpSz(36rmJv)t)RV>Y_#bj=I%Ac zOCdB_iGIBLR*fNEKAEwdTD|3*qU*wsN@5GJyv@_Q;|8?-^2Qy>p~i_!WfWO&VSJ5z zx5g|2uI(n89X|mS_hqSK_ucLrU6y*arR@xU7c~V^j0cNdhAbdq6WBJqz#lMEV4Bm~%|K^YaXb(m( z#Nq_Z|Ls4R%_oP1b4LB<4<3)t1=RM$SiqtMvTilfKuCm=U$G8;bL2X2%ev)tWpimO zMvYUckwdD8&E1(@L(Qf!erRYR=Rglg96W!`J1ye059PbmWw-xmS(a+V(L1JQ3qOeQ zpA$A@8xEZH1wWp>6Y}OVIh?LSU>db=dTBdpNK9EBihXqakjti`mb`{qE&`5B9Mld) zY#9ODFZ60f7KW*C14?D;;r(17OgE~l_kC=Bk^+9Rlk{^tU3<^sDd%MmYc{`}v8hrl zOniyZY7fyxDwh1zMj--pN^`Oh6AF|+2S4kurKQn- zAKd-$IQeF@ogCwu_!p=?eP-;l-Zj5hA10p*26b17^-u zKF>Dv^+E!NwZ~NojVnCo^2WLIXqO}v4SUk?arOf<-wuCDf1_%+rgNHL6n)Xq_oc;` z+U&%-X56fa%#QA$l~|s~TIuWF1WMcDoN~AE?XB|*KE`ezuJYPEjF5ijq*c*n^97?y zlQ!BwdiVgDlZ@cQ-CU7`mK$=HjD+T8SejmQGq@%jIg(d~rR_IE-i|pL=Kgat`W~kT zv5&+f_F;>Mf=URpp-=Ao+B{2xfND0HHqtDjJpcUa0Sz4;dAxK!b=b78hXL2FvdWCI z1O72;a81w>6`W>R+0Q50zXW<$QvMU>u^`8kQXHppVHvBVEYph!NAR&fRwwI>`AjWb z(5KE>3Aw|Aj*~0)B3g=K4C?5G+@uHg&}b_6vue-+G`mRbX52KHm3|g$oG7~N?at6p z$!kU*dyI)202uui&t&Q`0WD5AxbtIUV9AL`f9GGUMgX&cA#MRqNvTC=KC_*3p6vDo z``AT7zXb^|ZtPU&dl|QIew!$o;=!0d0kNYRy8W6bd}lVdo5s)b4-_{Yi0*o^IoyI@ zI*G(D4E*f}XF9+HfG$2uy4E*?MSO4m3U%Vbnj=B1AWoUkJhsf{*2OpCEKxUx1V3Tsj1#?bnQ=)mM6Y;Wn*;>W&mRXURZa; zh4&gvR1Csu+#GB3tqA=-VjIuDrFnSnw5tT4R3?wKSy}wQX1a)GQ zZE9^JT^*oCVC@i_bTj_WQWO93GdC7nTi?=Z^s2HP1bME%JkjKDbAEwQ`OF)}@pOeC zMPfvGo6b`UbG&i==nFD?uzg}R#bn#^9Rj{`@WsVrd90r1P8p|&cTvi;@KFg4o^^;a zqgAeJcw-?382S1Ch2lJ*pBOL4Op|JVh6f;HXjFtp!rxNfuE?&T8fpF1%f{Q8KC zSbV#0{CHwI;<=ev)e=9rt$1$GQtm+1W3?_&9WSo+?WZ#}7oEZW+T$vSg^u<2`^ulBrX1PM&|l*b|EIaGkR-wdn#%L z6@F>Vx;Ao0!X(pvd%m%%lO0b?^zAQO-Ooq%nMw6=$q!MiIgWvo2eXeWU+SN`*S;Ot zADQ=7IC9`9i!Z#h6IwDtZsRF0a)#tx)9ufhDNt9#=TEG8ljU1=&pp9hej%Yb$pqGE+jdDmQZvUYEC%QGV3Af-y~J4m421?1 z5N!&eS#a0v$X!*9*PE#Bl9UG5+YcmjwN0<*uz#gA)OJbhKJ7i%t9Z;XOftsRIkR;cn@PJuOh`8+!g{+&|Mlcz8YLE7 zLFGOOpE$ zlg^v%fTt=Bgid<12~_&(-$-DqKVHxd^X2#p=2_Ia6N<0T549x?*>3Ak>(!}gX4RV< zLY>rS7w*AR1k*>Ww2c6eWKQH zxN%$mwh%S1uN_iU?b{o^>H4ihGQ}yLZJzXI;(A%Ek`~bxCm!@N6rNn!n9fdfOHjiF z5l8|55+MD5`!&18yQemJnff%eL(*3`c7opHCXH8zJ)b!UYm18ruYs?D5SD@itMB?D zdXN4!vhW^{6E2X0cSBKncFPY@Mh--NmUWx>s>y}@sVRsd4t7! z$KJIAAMoy=b}Z%ne!NVMlSd>6scsu9&*Rcwzq0I*wO`6O*-+4~mY!UZh6-nQz ziBc`+)r-0CCiNY_5n%Bc_BBjH((R_>`=emC?fp?-J1QSza%NYt?V2>;E1BAi_lr(3 zzLSk&F&zstAH3kKbUo3^jizj1&i(vuBIU^CgfMdEU`MecREhxTp$^O&5{^rVl&)Q3 z{mMIHC`4Pt*uEYStj8uy_T($Kjr`1e&1G!j!OXEEcMQho-ivN?P1$_lhqr3#vjAd7 z+y70VSzWwDm`soqx3fLT)H3UP6e3a89Tp)^Yqe(v&T-JhzVIYGvI#|$qkE6HA?d{M zOk%hAJ^$V)W5ft9JLQnK(zN^T6;Je+0)&U)sgVJvlhJv@I}2DQsmoT534@<0Fd-iw zK7?3uyf<9xu32Apr$b$@1sW#^nL!tdEwj)vS8%Q+M|@CQQuH(Dv@YY0fVr_vh&uv;$Rdm&dp(K z#;(-#mfcpoKnr{x6~O`cf5ri;MA2Z@y%!<>p^2g>&AX;Jzmja^!Keq$kz_=3H=n&8 zx@#kHi&h?tp>Rl%4l!KegPGyO?vCe7FQ0y|cgM&Ke4)_e{-<$q)tIaj*`2Q%>4hG; zY^*AW$HgN;z!IN4Kn9g|cpNvtuJMcFUl^W`6iT#jCx=23HHe0l^VG*UAsZ_vMygT_ zK5QE&c}>Z?koBqLcP%n|6<_RUZe908@m!`Ia7@SDmEA4kc6uf6eWE3k5na-Mto&7s zcQJ42y1Yjn?3bh96C-0_5`WT%kqgY{cr zz4s286ZH4pZ>_m^X02KC$6aeC+2^e6efHV=-Tisr=gn(Pbwx7b8^jTsj8y@L_r`Z2xJ3-LP1dA9tpBR0dXkE2AhEW*g$Q75D5j+ zBY}Xnxy)^Ind9T*-n+5xc1im(rz#U}+A7tYXf%rpI z{2>nhP^drD#vdB#uM*;Kg9-nGWD^YqMMG_(Z6c#>P|=Y2ZMcIH>f50KuUrJ)|eS zuBR@fr><_u1~g;?9f|~k%@E3F2pu_uMh~GeLm2cB27``-q9bk4C?G^)r2yo(ywGS2 z8iSb#QJF}0m`D$qNKc=r3z?_`?sY4X(3MD=6%-I6S5T-GbmR&exT9Clm=z3a1%qC} zU@#~f3@Q>R9orzFETCoB@-SZGdX{{s3aQzx64czTbcp^Y%ERwLNPYH6Gm1*DM27^e zxiWmOsnz~98hlfOnW#&EC@N8oa%;P8;@oj<4ZXg$KH0m`>*Zy;xqB=-K88+=pR|<` zFW*H?QG=4cJHl{4noPL(ZrA_+_5VyCd_>{7zyd%yrM&$;Tw=riV$9VCDfT4hHIFV>eif*7+9QrophsaZ_-Q z$x(Tpft?}10+3W%ag=pzi6mMPz9jh-zvF@VM=7v;-!(BMv6HWJ*RK1%E0HZ5pyP~a z>UV!h78Twg=z|-Q!sZm|v=y~k(~^ea4w%3YpMD!No>zahM--%FJ#|G}@m$K~&u zbQ}MPccwb`)t=^uY}cK2PCtMz!4}AMZ>~pdrVWsU@8s<{&O2l%i~M|Rc2)RBH@4D zy!p+Pi8^03v03gvtXmgl*$a?2@_me`g}0W2JUCGGB4eEal%LtOllkP=xVB?!_h6j%pz z7D0RP6MFA&PVCRefwmH~rFe-^8+Z_QdEo-nhsh#HX!iXg(-(xq(fnyZ3@*Kap|{Q4IJ@E-I9G6`6kcUySc}hUmnRsa;G>!om$eK!;mS znswLqnD7Kd6PxdmKmS6=K<&yZ_#F2uK8Nw?Wr4KvAtugCqH$S)zX1FKCc-4s_WAuL9wqF6`)Dz1}4B z6@OV4|0!rx;Xhw&b}kgIT(~8#N@U{&u|MAY+YjH!Dk?DcMd@}l(Umo9pZ{8a35vza zvh6ANr;ft9WH(G_Yb$+=tO+4$K+YTag;77)Y#npX7HM|dxtvBcJgwMN( z&N^#+96Ar%8qH+LpL2XD`IB>`YMSICta4gxr?Id}<}~316JY*j@n&>Yv&J@Ra1OkSF;@Z%7kIj6*T(tSOcoyb>8;{?sR#ql#?u53 z>39#h7|NrAO@HYLC8DIYg6_m3;NnFnt+RQOF zy{Qb1drvFG^zu5CtomSQb78WOXYei5eaJIaqnU9D-Yv`-jBpJfXhoC1xVA)0$zFnI zTxKAAdZPJ+XDkqodjRRD6}Y0LEtf&uO)$ldYYfbgt2 z&dCf6H1qy^M}TmPr)IzulU}0wQSWa@#~d&5`SgCvh1z+)D_%3Y5#xSwE%h{L zZcveh=Zd;VpC6kH z8%LZ}9Uq$Yee9SpY^rI9#I~#U}lEg8{czn5kv&8RKlt-l`IMszdtL= z?2AroVZ?uJ;!R(~@`72mBSY5AJ|7=p=oMT~e2bt8%@T4w zouC5@L^N5jtD)lq+z-q}5mfvw(V>`9adK7oiJakds)C>lW!H#Xwx(uApy+V`Vcog! zNA)v?G)mvw>QWq|Rww#9mK|M*14OJP!av`)H?d0Ra?+3u!QXboSk&wC;Du;&KE_j@ z?zu3t&=n;?YZ_$@Zj=!p6Eu&iP@zXocZ+Q_L}_b`-N3a0 z=S20er-1d;k z%vS$|!Om_-Q-Iu`-7)T}gm88Go2H0QE#V|4{$Rp2@;}I=?mi84;PJt79lJ^jvXbYb z3Nuy}ZRCFS!eN>B+0tVdzGK~`%i6Arwc9)vPaQSb9rK<^uqP;c)kM*gO^7K7j$$X6 z7qjfiz|55Dj0_~}=(|cHQ@bLvSY1gpK*-7F5JMHxGTOnZ@6XU>7E~XV#I^H3FKym- zSyuDdd5jwc5SJDLCSIq)*(D~@jW{)_x^~Jyy5_3HF4tZ?pn;WRGb*Uu643PI&ITg z^QO#+f&YEX{cB zHE`2G8@^|s@97Kk+`FSYBfUq0$l$t$IN2>_aSa&=z^FX=BFh;Fd3e{PaPHD8OP-3jIULl_B$KB4^@z_dV&HeTIapGxVKM~mZ?M8`LT}ZoW`yK*Yiq>*3FD4&DOad zf+D#+hZx)eTR6cz_^Vvb4?H!Fmbz&JdUpTB=@Fc9_07u#cUp1s{afP$;hTk8Sg74L zk_ zy08GdVx$;DUTra5u>a1P!Ym34RkXmcZg9YzYHKW}l)|a)v<%RyHpNCN zSd6PHvjshGdZ(2ufL|4gML}$(6D{}6sa?n{f))0q zBzq&ysh{GM_AOb38R=+QBtTphR3f|mpnnzSbM~y9md#DNb2;p*Gb9*@C6o5S zEt2;EUVegonOxo5_Sz^<{!-Z2Z~%FhiS8`G+4Vy6@&$;+-$82VUV3@C1lls!DE# zPC!Poe?=voR>GRYf&TmRl%8)D$o#J;5$@S{I3V30g-h%K&;zvpQq<(X?I~DT{Oh6o zPjk}P`~Lt-s+ao;77zb3%*nz0+;9EPfHfF zhP~r}-XpP71uBxmPK{3sSJlNB64YKb;fdr6;D-UCF*6d3Z;;os4c4n|Ry&P9!o8hl zkI*sz0T6$NJJJ)Ddv}w3)8?*QVnvz&F&Yq;pSoe9yI2BRUj|+&40+mqceA9D4@G4WYoi`xA6n+F8RR(IdyF^SsK575uZ5$+dwv}3FX8;70_WHnk4TL-aBcuI zqPM~L&d%o1gBg0~f-kR7#bK+23?GO#P3I<+KKG-tRqRfGNC0ZGuO^~}9xVO<9d$cr zz2LV9R$M@}@!N6e7JXnTdLJD^J8-c7gdSKP{mrjAp@>0X8D%)jgxQ}GO9B(kD0lcQ zxC28-yjIVp{&{k@inPlGx8e7ThDE_@XiTt>uoB78Rqv;g-j7pE$XQYe5IuZ8xSlQM zq(!O<>7?;~_<rNBMv+q1F(J5Y=e$9x( zJVG5=r=3S7JK!p3Yx?m9y?5FjTvkFCYn$FdRy0QGT(e-K=rO1fTjnkAAv_m!$J2^{ z%8743(;d1i<@8qx$VYy?#ynahS8J&M1r+hvaJ<1x-Eu`!w2+9AG0_26*pb569*G_< z1v{Gy(cSULy)nkS_gQ4YDxE7)c7WkLCk^*@hBG~6;6TCss?}>=3rDdZ+y)`GUwVW~ zc$^Ci-lGH({eH$d=E=tX8e{SeEhhMEV5=Lcw{&Zqwp7U3Tw<~0*NEkx9bvbwc^rp> zZtnW&ZRWt4l}%^gpz!cVDYGx63V7Ynw8QD^j*B+<#sO$0Rh#TS z^*7Si!jmEXcoisoINPZ z)$FPW;FwGgk>|v>zZu3+5}y8{zIMPVNmJkM)Cel?5vqFC58sTvB+mLFV?gaZU#bT1 zf~aYF$bI}#mtggeQi0=XqF=O-=!WIBRP}-iUw)c|g86PHh*Hoy$^nZxHBm>L`2y5phvSHOW?IG0$Jfv~W_9yi$ zQ={t04PZ~}X?M9>TDnla{e0U8~!`(Mylh+F;Xp6Bvl-4 zz?F!H6twMP%RA%fRCco^+ z4(9MUaXaitl}ERm%U9j;%%v1YyeRF`rK=olh2xJJ{464SB^?fz`fmFM*U0OBJT8arm3`zVV@QwqX0&;t z#5Y@8vRzp9uHJD(xKl|d<%3Mz$nZPgsr1T^`@+^aY{6FPBTUyx_yZVFk-NjC&z5}^ zAHG&Do1?ELf^wC_oR-A+S49LvM#O(wW6$My{!9zu;|(Gc%mXJL^t#pW*iDLUUEm4U z98$~e)&nc8G^vWnGM*0jL5H7An2z_cbr4jP=`6j2YyxP>aj0Y?@8hlGB_6Yr7YA~zc7x8Z5G z;7HBPUB3ETU-9|RKE1!4^^U}3BhU#}Jb6z!sz}>g_9dCo*}y2Wib@Pw**N!x)n27d z@C!4L`(kP4+vKn0ueS-Eyj4)i9UH7pYKnpljMzuHub~wTV1|ikXWNoTB54}6AHLkc zrrV)2aZ@a|1;!Bx1G!Af=A>9PE}!P)dze}sV+Rv5ZYsvv_OrW4_M=RlT3TSi-mGW& zsJBGwi7Bbiepq}X{Ui&3NF?@Ph6a;#vmJG!B!oFCImhKk1w86!EwI9uo5u|7#vVt{ zF;Ph@wmQM&O|$R@*mDs#{~pVWxV=;H&WqW#6>LdrYbFH1@$4yfhk{3WN)|a-WrLPn z7FmzCz2X-qJDRrkuA_HghDf5m^%HzYoj7SC(b+chcZCDogU|f(*qe-g$QJ7_P*MvH z6}LtEEJ*>|ClQ(s(0QeE6VgXg{g;Pf0x1sP;9hw6RO7LwQU}FI;IMM@LouI*Thv%v z7XX&+^K569u3K(I*kO!Tb3HKIxFgS#F6PW&NcQ$Er7%KR)Z(1%20BB)?iM+R0Ol?ZQK-dR&L&mwl{>98P z31ES{FHB^@VnOj+%<1JJ9Muld!N~8nm1c{247d_3+eHLa?n^b zof!AjJ$PiMvqId>fQ^4n=wf_l<1g2@9|4tLTJF3R?9hXFB5n)+%0TQ~kf1`vbsNBa zcWQjr>+K!1V0fSV;Snh7)SBg8$+%GoJBbvFr|aT3{BaiAAUP_dIWcJU}mmi1>D5_xB(w zGF5ngl;R9crnA?kNW+JC?CJhK&eySnZ7aXj@&qNpQcZN#d*9#=B9X@X=@>(j4J(0C znf3|l*i-HVDf=>JhYp$URT$auxjs1($;jHud)#%8e2D{+;hG56m@}t*+f>l4e1tZW z+1iM-g+{_rlcK^QM08fW49C4pM@gVKpmRc>i)@16dzI;8{hmcGb5&5V(c$ zd}ri6l5zQ_$!1dZDo4u`z{=>er#LN!TO|0QY$eR_>UG)yDg*okSQ6R{st*^5kJ+1DB+7^Unb60ajIBzsTMdv6%0O1TQumZs@_+lmN`MJ$oJCc{_i8A`)GK8)_Fc`5Ld@9`Fxu(qxX?bM{p$WO8*YTqr9Gy!p0?H4JtEt>&Fi zIez7U8`3F}UK|FFdk9!_@yU<6%E~}0OVtfU-47CMQ^%A*y~Iw#^jPD2Ay16!r3KnC zMclw|#<%_z9g&o{z5PQ*XLH}@CqH*{UlqzkcfB?|9A3F5W;}rdP!&H*+nX*mtSr4) z>3Iy|zSuJBUvn}uqu*43MYE(&a{;XI2S5&85-QA$S#Kt_a1;dp+L@V>u1m$>b~Z9B zFHkW!$S#KyylsQu65LA$RKxF6weKs2Dt>*JB)%kWI2iMvS5kT_e$E!%3NvrKN-h@8 zoObQ{$vfZxDcL8jgYA)xVZnGX_uYHh3$(g7z4-R}848nmMB^eUud}8KF~_tG zJAHcZ_0ksp+-K#ru3co7w_PxPdsdMV9zOplqQL3?Z&jHRui;D1<B~M0Wh(_y&x>F&+PWJLtn!QLQAVcFOW-J@Cn1Z}9L{(O7*?+*X&tPe zYLu#S;pT>8(NZN7SFo+8Ta40Q9*bYcR!qC{FO>nDI#>7r#)5XIpuc=^>W;469|EGP zURH|ncWXcFcH#jwG97~@CC+b+GUV#=Y_j;0Q8y8i)++)y$bg=4;CbVpNR6~#L_vC_ zV~%;0ngl?{{g(vO{|SHgzu~Na?Ic24^C?g5mw!=7C9$Ss^W2WWzKtd2fE^p6I3)FH zff38AeZ{x-Qf+urCL4+!{}NfrQ%><)q4T!Xz|;ZQ7OWnaM9g@XO)8HCn5Bda6>toU zE{Kj%*IQb?=CbQ@?;V4utFO~GRuz#3oA*5j76AKs2bDy=lw!C6x=h6i%OJ`IWV?ey z4}8lPAhSc0*s9X*4%Q(Vr%pJWv9#np&D7<(rci-p(S`bgjDUv#`=I!Pg;N+r2#`5Z zeVkQ~2U7rvE^P$|-)doV0Hxt^Non|MVJVFdPNqasWWH-45M2(|7r@M(wr-4XE;qCZ zn~i&HoXOYFyPo?VU_nT5HZwDZt@wlvTMlR03aJ9Z0MwXK?&~%)K+(TkTeJyMfnShu z;GF-7=H#}W{`}F)`X6D0X6wW9^z6O>WEY^9GXFe{x_UkB_P=X#+qC=uSBafhP>F%2 z88ZHVs>oouHbcj^I z8e#+*-;GneJb8Jxj;Al|z8sz`o_`kb-^E;P?NVSibIKK+)MUEXj=y^8_}98C2L7t$ z?Ls*G-o%Xz%ab1h&}?x+YCcLX=b0|P3^(_dZG&)fT?+10`@qg@L^gqW?;43}vcQWV zu(!#f25!l9tZtZh4f46bwZxoJg{W}ttdq|>(9sNTZLK%7>zYego*&{@{=JgTq}}J% z^9`-OLOFmO`mF*>#CziQL)<_K-pD?bsxPxW>15A9jdIs>RSOit>3 zEcngmuIeFW-Tk7B_Fw`RG`kN9odrZmz9K#_pS^k2l;7S>{gJK=Y_4gZ>=-i+JqjIk zy5@fo4$n!vW#T3>4GArkk7++C7cLwo{+yKjnWx_C>lWG?@4&Im9d83MV!HqNtZMLh zd}p|M8o{=@AG{pe>B*(AwMwtP;I}UbC?qO$l@Q@dHz`$y)sYMiT)oQQHJmx-gLr{Q znWa#^)#KsjxS{eVC+NlLk7w@-zyxvyu4_Rr_&(qHL3q4(-xbEZbVp1 zTd&Q%89)cC5GRdD#aMRbY{Zgouc9Ib!b1NlkD8OgGXMcE-ecgq9gLZs6zZg>|`PY(c9lK3!9^u-M+qe zM|O-%hnZ=HCQO2>+85p8GCR?}4=R(+j|ao7h$wb3$?Ab0p{g3ve*MR;oN?^~w0pX8 zKT?mEUph#&qP*_opClCOojr*}#;3Z!$)4aG4^L zsn5US@XO|sS9Y|_qvl<69fcW*_5l$k$lCiiD@&;t&Xt%J@98Q19jn*mutP@+z7CCE zVV2{Rz*Y^}J-*nXIosUx>i3>8H|a^Bq;wr1H1}Wl+G;-i3;md zVhr;^W~^Jmg7Jolhvn~&IvM8o%1@PZwp(O;O_vuIS6mT`p{X)n7sp#(zt0dl^^XP{ zz1ha;Fb0ot8<9lxh_Bn%GiPpk*XoFC$0~sR3}e5&uctO1 ztl@o@tx2BZ*b?5Ip59bGDQl(Q&~zbNdfG=ovp|wp9re{6qA_FZj{9z~Jxsw%*lj$5 z|BY%uZo_-`tKeF!1>@r;!9o3_Xg_gv9}8nWh3+XDvtr@^mkkRqNKHXi_6ZX+F{22i zo~d=;^X$Bku+N6Ut4h0rQ|5~C!>>H?Px0RQ7VlbiJlTEWNu*PBZPcE3*-CXd^~u&5 z%)w5rmN?u1(6Fl$=qYImwo8|73N>dwG4b6M(wO=Dw&}^kkGQ|atJ9liqo26&NxIyH z#ofg9ykQYk2IIuY(ImkX>znJz__oa&7^f|tjXBtAkH57h^tIx&hS|q3DodJ&4&ydNHtbjOuByT-jmePe9!qnC zQYv$2jebU}(b_99g>s?l3$0P+!M#$8Y$byqKIF2iOsQ8piX=!{Y+8PCJ|6WbpX(xG zq;7T}Lf@a^citNs772)G0~2uUVpQI&;DsBM2T&k#BW6^;_*N+sop}cG9fNb!f3pm} znuh{-vWXJewU0xvdzBz?dzJ2MbM}DFtFl3&Mf^wEcOL$YSX5z;0FDsSpMRhC>R1L#JPi45#4F7rTaL!$JFo72PDIMwD$NlY#RDspXCkWWYgb?j$pVcK(_C#1! z8tURGYkLP^-GJTq<8h`29y#}&c#9A8`*+4fRupbGa_VI}^K!Te;ob?bQ{Oi#@aPL4 z)y6d~tu-g&r1ZbevhTTLSY4^h?Utl;$&^Wn2mL(!`KHGF$Y$$1;L$?WLDbEWFV~sc z<>;sgIO7;~pV03HezM!MCNAe%A}-L*ufaha$2{Yv{w`1>5?I;XxIBGP?t7i?WY237F? zNNqphzO|dKl_wkK-i7P`peJ5G8qex3Q*SBa-2S}aH+ymii$Ziz&@Ulw(af<^4S_ct zy?cA>?6HEFRJ}&6Q<<}~vY?dX*C>3@vzS{1D|f^hmYpiSf9V%=TT=Ssd*Wi*wFJ%V zxzFud!P!RXrdN=caRXm`a=om*ALmWYBvEfJ(Hl*$&|??f&dyc%GF$1e!UM}19I*U- z-SqDEBjK7gGIf0)8Mk{9w7#Y;d=)TO!TTTFbNR-2pM4(*(^LQJwiA4Hl$Mb68Wz^& zZxR$G`D@^*lQlf2`OC0uW;K<|vQvNk%f=jP!=KXw%+LB*?e%UpsU!RHcWM8)bqURO z%;)Z}%)Hd!SbsTmsaXL4{3{#aMk4IM{Pppmli=f?zY2-3TVWv?fzAoO*E#@pu~t)N z(h%(jY#D8pW}hwp@!cW`E`NUu>DZ9m4_NpAfLG#Vz-!DAiw-L&tLMJ-|5ML$^Pccl z?l78&>@TbZ=#hk1EQE#7m-Da}f8hV6L)WAX#zUCx%KqAspucc0Pne}@2SN2R`CrPwKji{E}-(l zOghy+yXf!pL$dyM2$r@Xdr|dfV_KpEaKk#mHsvRzBWh}T4OXl0HT>iB}p3RUo+3!h9CckpU$};u=J0Lb`^|c0B~K>5)N=0@_$qWz#=Ag z2Mmq*$6ag#OSV+V;TXsR7ST_%nV$d+0&mV)-@M zUss7HZ&FUc`oP1Z^)U9Re4UAhDjJm#>$Ud95IeY7?=iz)8a43`VGg37k>+P7!EPn? zI{uetg3$buZ>ielr{~0nvnu5?;n7nwJkggk3D}=YIJUS4=S0Uu@-J-)So8eDnINh$ zu$PzJ4%YS43Sf?6j~*@HP4AW^E(2;U)`zs!5^<(na!gtNp$T-1wX+9btT`u$$}7CB zW0S(c>9)G9MPn09&pfU4M0p=K#_)Htp51d3Wdj}LcdK*)TNMS7-aQ<@2j*DYMwSR~ zkLt`LN1kB~a|lE9-9NsxZ^~}Pw5+_maO2y90_F`$r;mU(^}lh3wY;;LSjdZYYjq;3 zY+}uOMXyYhR+6pLAgg^%fKsA%N3YAn8p>_R~NEijsR7YYLD~8(#u8JeT)sJfZG$MfikHVd|-lIl_t?4$qrVOaEPxK_#I7 z%TD+oX|8!=I<@+Chl{I{BfhVZ6KGT>hx#AfCI?RVvBU9R#k=C_c`|D%&bDl!9Sf!c zZD_xoPcDrB@47bq;z}F3x8oLPqM0Jz)a#+0(qjmra1% zP6>vgzG~SA9M?~b2J#+k*0+r0MAJ!aDxnQKn)@c}H$8SKRxt`z$n(?9jaA+_FydFt zRf!A!y|5aWShf|72{mN2N-VF&m!Q;IwgC*&X^Xnf5Nm^Frl8jom(OLy;J(`v)rqmW zj4scF9Xn`t5tdN>b~Ot1`|&H|eO-XlXBF)tPsh9gYg_^A*4rW!wX>fh^aL%mVl*sf6hbnjRO@4VE$dAbyUtuV(+s@UAjyr;Z8;o|}EQRDvQLr(@r$H^`P$lfZ zMG@xk!sS!%fp$c^x*bs?v2bQsReDANnCsEPCeS(`Rk7a_9fH2NU8_MCbCk!4YdhCPw&A&p|bzoubB;@9qozp6r~%{h>NGZ6|g;Kk)xF(LEAGA5edufF;x|qGOjoOGSp}(IG)>|m)34%g+?E7SOU%oV z7n>8Kyh6RsySe}uW*)?8^3TB3@It8AirhX8a&|3Otr~Ep4mu<;pXF z=`LuOic~7q$tI<5U;g}+j>K!i$M*oCxBw5R$KCav*QawDS6mBj?UmVR7wTQVL1EPg z)~d>Ueqxl;#LzPLJD6yJC`nr6o!yoWRh)x|?B%Q6{qlJkcegmUH~d_m@<;C6MPLp? zX6YWty0z`TJ4FUjO-%;!NQ}|rYoC~FhfKaAPcD}83q^Q9eVSD)J5@9b>|H|}LA#;M zf%df5l#or69FP*%@7r)I8288-+@SS@!8Yjpw>M^ci<`qX!|o_+eF-+O0r&}SaiapX zUQlE=Je3l6U4C_s6zCm*4oSO}FuD+z-zIANOa0fq7aGrtF#6VBps zAKfVcU(|E>FtC;I0Ua*{KBY)jjMDITvs*tMD1f7+Q9ndTwE(;8$R+*}&>>s2gC zQX*V=X1%>HIP_MD;AK6bF@w*WYQKY7My#}2r`+wiv>$rh$&r1k_w($NQA_x-?`Z=Oy65`AYYWmY{X zxYM?o^p+t(mt;siS4V;m!nkw<*SyCZ$isM)+mN&kS)F`=8x%fgYi0lUJ%DkyLPnnl z9i@AV(L!A^@18-@`#buV{BkT9bwdb#FMI*dZ zib;lueoa$_(B2SqbXN=fK+7b4s#ZeP1NU@(j}l48{*i0z#buBE_x?-TM-0XHLOazf U_TBc_zlAND070>O+yDRo diff --git a/docs/README-unnamed-chunk-33-1.png b/docs/README-unnamed-chunk-33-1.png index 7a1e34cf9780d864520f58a23f792eb0afffd41d..b2ff1a4a5046bf7e6e5626b7c5dbbbdf506323fc 100644 GIT binary patch literal 19126 zcmeIacT`l%(aObQuF$o2fU2?#;VtT0SXfwuarWZ*7gh$8m@rIO7zQY>QdL#UOt8vKgv-K! zugTx_Wxxjm{FPyV@~Z3#Q+9P#>Ecr9;n5r(-WnU*pO=RT6UKy@V9J1R*fk&k3gG*5 z!CzFY-M%qv9KuPum9bU zb3{dAVZpKFo+30{zHB61pvL*#rYgrAOVso3vN;LlytF)WC_Z9mQOX`l6dT-?`(*5@ ztaewg+Bv~9_T=g+2s$MZ2J*v#guzcv&J9DOrcu#L4bgB2vX>kSevi5e3kwE?+{A*w zKv*z11Pd+!!;*Lp$9j*20Ok#eg>@5j0}J$j0sbfW0EZw(zQ`aWzy%J8uq8|3;tr|z`` z%u=rQGk=)@g4KB~yZVqekLFC;=w-e#mJci#{$_Lh<$&%>d~IYh!F>xv*cR1*GXVto zFxGeA-q(64is?_*S^AU&P;qaN1MnY7!Hp_bnh!QM(N+;Cq>g~~RI4@-O$P8KFro+~GG12M0ZYS5p z6Q1PpnU1EEmJC_=ZGuNbJ`gUe!9q57s02FD+}3~NKgzd$uoc?JU_X>Am*eFX zJDIXERvN|eTC?v{hQ`--9t_6nNl&hc3qc;s(%ha6VYqDMSg6i=V8w@cYlbCDy(&)r zbIUG`iO$0%R#_@QJRK_M;jnLBePScW;t>Vi4lV!4(axrFsj=jHL}!(#p%L{6>l@KC zn7oSVD^t%;kJg`@M6PeMZ&hlW84mOrGaiM$O7DsMB?0@@obx%&lJ^qjAF0`VrG)J1 z!wTacV8^4>k?OyUIp}Ztx`OjkaS1}-EIvCx$nk0So2Xor@29g+esL#_C{lzh8~)G6 ziqR&xcr4j3wec(Z+1ckWS4#q_?Crj?evNo1NbXER-3a6NBHo*wa!49UbvY? z^0xKPT+u+WD7`)zdVWfWpp*{_Vv~o4uh;V0kfNER=NaAXJlJSr-!Om>u8%UG!B&|F z-?W=QM~0Pc);UYC=M4*1b6Qi{7gtwuMh=-Tw3*#ToA1OW@sRS8G}rd*Xt@#SkfBcl z%RHqLxfeRx5;F;aDW1QLs1rOnHOs%L+nLshRK3rJRE0Y>IZiqjw{FLapD%&MH!e?h zlC~Zn@9}$@-=VjD?x(7t9d`V1Er;;i;jN#W1L73o?9CSXeubR{hJ6Rq)1!x{qS;TU zcxroUx|xc;-bmgadI6EVdq${obaV^Z$0YQ55JYyhDKho|QW{Rq&Ci;GC)vmO%OdQx zK>k;9&J#;bC#i>e4Z=imRGpjxM{BA`4&q2=hj5%%aWe5ZDO!DZTX2~K5wejb;|#Mp2hKt7*6@<_-gsXu?aT7% z2}O7a=QC)Q!2KUT^ok79zAy98)K14B)0O^wEmc(QHzD1Hx+~*&4;I{20rc^${)wLN zkX$o!G^wm^auZ(COLMx3dyv8KsDeAF-Zc0lQ_WTX*$f8H+AS{k%{o`QdTR1PfmAo| zDA6z7NpOumxm-qp9Cbc?$vB0YH}dCyh>`E=qc5Om!Ye6N>+^VXzP`4-vmj3j?9ss# z+6*l27Mc!U_lY7uM|Ew0&??rKv8Rh zb#B=bPuqOTH!T#kCUZuPTooB1_n~KwU5pHthLTH9(LKR zA|m8(lT7qtNzfk^hJYovY@8$vugvE!l6-Vg-daB33EZ5p^0t+xLI3su&2Vb^K2&bY z?HfJ`aTfnh7gUgjIx~>=3vw}SSnDryREJ2m!6B-~_KDVfd!2{oDJ2kd;I~4wn7&QC zu?$RfhB1{Gl=aR%Zk(GjJNX{OEIUVv9hA2Ohv?gq_Ck<4;fMAuoR8$9^)m`yuHygP zFpG!JE1bNHB2)V!H93u({w6+n#)eE*@W8i?SH$@ay^=beH5ljF4{dD2SAmW1a}kzU zze`MprxQ^+4WD3!@P0;Wo);Uo98g6Bdd~6e!q5^ph{pVa+G3^PD772WnjW4dFsMEcM%K_N$r7@lfyiG%)d|yKbZU4vEpnix_&DJ?(l>*#%Ag+gf{Pa%BK>g+X>W_ zMLCVaReP-o@Fx1HgF!w;(14h86_)Wqj&o$5fCV`d?)Bx2KlsDgICP8Om%1Hh-%0?7? zB-y{Yn3@p@_g3rTpa-SA9}MC<0>;{dork%%Y+iN{BOc87+sja>0)ZYxaF zM=J9vOn3w!0Fe~uca+`GE$-a?>@Z}->zuZjy>K}aA_=a3X!-cuflcKB&o@CDzN50N zhVm)lw(rak+_N3$w(K9N>Hyo<%fjkLn$l;t2fpZjLhUv+C=+@eH%OH&=;D@K9E-fs zH0)&~Mnj!+M?8)WzKRyLrO$E=yu@VV+BC^GD@WG6uA;AZ)gsqZMNE|4spi+ImP=K? zC*b4@PXP zvgneW3PAG7Z9Z|ocMxu~mSw;~NIeJAGaK*++(o3*1>3yb}&{UV$tW(*X# zIdBLe>zrl^ZvA;S3|o7X;vT4jZ(ZR+sSJh|BU|pw$8AnerGSB$e!W#1OlAb%;l368lVBv}ZfDViYFITI3)|{=TNWp|lrifqhlf=T!y- zFDvWkX|H^ehu6lkF!@a19Iu3_sRmWdyIpT03xh)vS03`0+(3jqYt~gFEN(#P&R4Oj zw2O2sm(Z&a(On(4VQ|@mJ@adqhO8`}+o7bpN*T2h_l&hs7E9vz z0X{?s*_#ZmmWAtEdbljL?-G&8pPFuQu94uw&;**|BM0X;cLoiUD36KJdhpy=54R5o z4M0X9WSC{-91(~Cp+M+0c!yfhqb_{}zA!RJeW-H_DgF22A8UoW-2@?o1PisW%b9Yl zQlbc;Fwn+-b|lNmBq_go<5otS0pa&<7ta<4>eAs~i`!RwbKcyYNhpYZ0RuVyh3ke0 zv#S&vxHJIuJI&J^A=xexzo&z)<};pnXXmQz2wU^QJW(f)^xCseR$36rCpc~;IoMk~6&ch! zfu_OVz+f^|%&stKtEA)aO4Jq8`d%srn2hh(VzsVe&`@_rxh!b6HM4h1w058`9zIaz zP5Uu;m8En^)C;j}gm?BO3)kJM8L?W8)dZK`NcepDSk^@68~e^2vPX_r*#GG)%S0|5NoGtDfkbbQ8T>Z zEIscCimoe`{A+epD6Lz?jpUtg@$sp4-h=AR-bQiiIrZHq23b`%@){vni3n#+gTJ-gK!|&5 zAuhTNe+h;7!(bBr`y$FTPu2Dz?^)(|Z>lg$L`vVxEl19B8HR2)gj{ z^Nq0Ab*P>!2;RSs=eEF5a#>OJ{D7} z!s_y#DInpD`-N!Fn;P@e1x$R$u3rE_?qnEuCA+#ZpRsryJOfI}2j7w^Z2Nmh%|Q?; zjI=5*Xw3|t^0qw7ZYaCd^IhJaOp_v|T_b|=@66#P%+G|vEQ|kHB0Y__5AA?iu4mVx z30A|u%|2!9l4x1ry8ZNMe=XDXiKxrh2+a|D$bV@T({-fkCsbeFPlHyGM{QZIrfJTV zDmoj~Gu!i*(JanshXTgLf9nosmzD>RLER5yr)hRM4P|N0Lw(O@CsM68kSYk(PFO|7 zlegVtO5wtF>#MeV_4T4cC#&8sTLeB4C3JIyK8MMhCs$}v*3!ss@pv4s@FeLCU94^` zz8rc##6t4yv%zaGwk*15$T1MZA$jnK%c*b3=PSq4#OC{IFKA4MU2KJ03WD^$% z>W^q8)t8Hi2-bZ*NUrptUeM-OvA7squyB4Eq@9KWbIu7+k@wCZ9U*M)%usV8GU*54 zT}+69L7dk?xgLxZ@zPvgj_A;_?GXC)vGs6D`W^9PK|XQ=@jA) zv(*Hc)o|45X zkba;-qhUkl%aMyM?7C!5EanpH+ub_xF|#H8mFEX9t5=GqUWF#QuZ2}#Q>C^%O_%N( zBz5EK5$q7Pz)$ZaE+F@-)5;C`@GIeGo$iN}pjH1sXKSa+tNy#f=6V`o#T<3F#_C)U z>ag#mspgmf__AnR)}VZv_SkZRstQ&3#4v9Z_{x6L9&*MEom5l<45mskmB*tbN7kL8 z9@dlMfzj}%en&~A`tp`Ih~8%wWqi2={F0P|uVq}B-|o#`{XVv8Ev8=gwliBU7Ea&g zXtYLG1cwBDBz|dX%mksfE3gZX_+rWgvG=;D6tz=Ltc-0zqHxV^L?#nAt?w3i5Rwt2 zeQ)iXl;bp(GVupXz#Lehz8slg^D~Crw5Mb+Q0G6|$|@4q&srS3waD#nvwscrjlvvI z#rxvBW{G6T9#x;o^|GQn47Z4%pN!BmEfXUfni_hf4xXMwF`P1g1lv53%2CcDK5|Sk zYgGcw&w(A|^3OR^4xFwVCv&lf}3eMNZ2JQz_h0?X%f$$}35RUJO7bWqN1AS>0Lmk=vf4mIIe< znMPU%lx3B+!&WMQgY(9!^~@#o3(CElCN>9_3SW;Gq&x{>@!daX6CgzNEN41msI@G4 z`j+2O%Sp39yzrD@mugz5+5&b0J*}&O8vD5!l7~`STF7MLBYv&ip0?4y+y~mkLQu*B zFmHljSe9jq6LB}g`X=SL3LBIIQkth@`A!;EMG_8=oHv|mYEYb)C~7yq3z>AnNd4K^ zCuZ=ZVC?0$fay262R=}#WV#RgwZ&Kc&nE51LV#cmKy=W7pG23cWGh7XGs#UpUhsuK z(nuDlZX|gRU#GhB``xTWYLyx{8OY}T-}CA0f=hwkee_BPI{8}{8V72?bSV9;l7pr) zm9W*Q??l>=)g{!@Xw;MhM{~5QO>Z2SHtRERxEXu8ADsdw+2hG1e!|8&v!*2POLD%E zp(u{2lR0IWHOGtcCndk-OIbdBxI17q>{g9?ZJA{+tmn*~7U!U@GL@bbY7?4l)1}cF zZ{_Zi?{0^sOc5i=_>t4nynlIdmr48w%C5GmqjS@Q@8DT3OQ&WvfwdPRAIo$Gm;O>1 zM&0Ey_m*CS7vV_~sl=Zya54$0D&BjS17~$zLX`)ki{|{7p@je-+|oTVRVy`*owD7( zzEHhgN{?t>5sf2{n~=?gP;qIGSQKrkUStE(x+4Vn7EtJ4xGE$mxA4gON(T%KbN zd*7aRC~6QPRUJ#rXR(A-)#{8{U*wg44*XlS5~XD-y&Heq6k>xIDZp)O@RE-R^fmA< zYx`R(dv052(gEIpF9|vhb${ylNfZP(eci|Yn(=sV z`0Yyc=aaTUTHOS}EGRS#g$qN6Tsu6^(%c}rqgIGS8J<{$s*6p#v!#08s;HvipLz0p zhPKJg8aN2-_%AG+5V2{F`N+qb$~0VpJ;p7_WCmjD+CJ^p17q#zP)>c1cW;${K5P3{ zJDhz@qOfbL2O`iH{&v5=%4&jRit6JRt3Upy-a~$s2&iL^*B#Zz8zKoJ1VDuFn=?5@ zDH)iLPXao&JWCsd!{!JqPBaAhq*{19QL4ALpl!enAI(*xvDZYGq-PR>&K0(a5~l{e&Mm9p zI0#jizhVre7h~HWT#}Hg1?!W#IEg$j?$(rqKxezXS!{mXS^sKU0ss1#JAZ=gNOlItp4T7Vq=b|9+{D#7hAk@)KE$Fo+z4%G>?uj6_!|0(UCYsW`@5Zom~zg zKNEes=N5)nmnfUx4eyjb^@nURhEg<**wCPzY zi=Li~ow2%y?BV;1_E{Btl|3-GBLF(7mR@5PCphhi<3~DKDU9LUYjlubX{n30@f=aThLxl6qmWKzi3e9*T9d+HjFWf=-kDJd29I_>y9YZ~Q z6X7*h8mQgr_wG35g04b(-iTYs&}{>u*x=$|F@NP`5j1|}-Y+R$K?&P!wd7xG`|Dtn9^t9S`g2W` z!GQUDntiDSE189Q8WZDwu;c9Hzp+MgLO0vaD$fOU2qw?fqEN-a_Eq~H?(q8`)4Xx; zXKVU9WjWa7_v{<*dhY)#)A^G=(Fwz`d{WhP#-2^Lg6D{UK=<6 z9r2Tx_7jbt(yVDs}$UK&oa`tM*+ZTzWw z&4Dy;vu?Qc5UG=a?u!~*8B5JE+7OlmKR(;ppWX=#^{DOdO9r-GAVza$8&O(t>$o`H z<3fml|JB7mxd11Dn7&qvf?{gR17PD=CIlJLTNfW{-Y{^A@at-4aX+u=zdK;4o2%hb ztpe>qT|!h~t1Ka6kEPil?J^hVxiW0?a0+y-ITyckXnX@sO8q(bwdH4JeQSAK@FWOv zp5|x^bJ10Tf*Qora`xGU-99Ta*6^OCC(&I=UCIPeqbJq+u@Ud78|tHkw+s377a8NC~qJ_V?h8blSR^j`ihKNw+5Tuf816~5i zh>TYBWPiu3>IWGj2C^0IJT<&?Mj#V+KYFfPioSMIk8X3~;|qaxYcf#BI~0Bu|imEM|o`8&Qy?I~>PQl9y0$Se~QNR_qw_@gd}!G#AH zTm)F#Eq)`v&(#mT#v3muwTDE~SLHpacuaDmr1EWixHL>!!mT50ieR;qp-F|g>BG$T zlgO<7S^HbuFiL1Nhs|ZJXO=%4t-R}kP-IxH8{fYasDz(iJn8=+w`dIV79>V@Itx$j zE$@`#6`HWzJgR%aD7n4!yOl1d3DG0C6EZnY;=^?-zbkeAVe7F1?MkYHRRvctJ?KU@ zutPhqVfq_~hHtaOz;j>(mg)cF(%Px+OC-Lc$c*5u*v_JjLt@n$E;SncY!%erd)0Lm zJkwLnwLG0Y6tb@PUCLJn%2Q>u zi_|8=_7m3^C=do)D2U`xC&?YtuRLUMozU{-7z;wHhg@@GHozWd%}{n53acH#Q=Hj!YTwK zV#cYk4&jgp5$qpyAftDtSR+qrZu~FGnQ6cRR>*_Eki-pzqN=^B#P2KRIPptCtW`(9 zEa=u)XQtglT@|C&GVx6b?;+dC{ghW9-K7KT6@d8)IqrcVly)v8DsfmCb0nU3}0ooJ38?tltKo zf62^(S#L<4u3GWd)Y-4s*)JbHt`5GcUmfRC;~Jg)EeHKITWl7$tOtu)nO-xu8+SNg zHXe$dPgknnkaKL%rrQnhTI1FmdN%C)3E4i0zQ-cOs(?7kFiW*Tv=w|Q&-^wp=yLf$ zR(+nuFwXtMk^R zV*4?zJZ{_UvC^k47GfE{``QaDO5lPH;k+ukRcN0e+k-+fe&ZyoZew^ z*g$h6%uB*(_&ur z%+<@u%>423NGuAjA@Uf>#b>Qjl0T{hJ6u#f_Q(Vi>2L13mG@|(e|l^eq2L1&Flm_r zWT8+u_M4-eh3|1E)2~fUkTB;%W6=>2<*Vwm{<9KRVUKW{iI)#j)3)#tZQe}G;b$!< zaP&9zlD6s>jGtIXT_`f$vVfYc zh>>BAM-x<2^WeDv|2ViTzm|zW^fX8ZR;Or6QpQw1Ur#d3v334>Hf6Tr0VFo<>-#p( zL}PN9My=C_enJ zKHG3+HChu9W(nB#q8*(G8(AcbRx|)}6;y2DTz*jr`%RD#Pf`!FLqBiJV>rMtcmyBaog8^ zetSn6X$kTQ5>$GFkQH-!`EElQTP-QSZ%5)?Kv*b+7{C3x{LG$cK{%YI0|`%Jp!j+h z!T|pL8;5~@xEWP*j*E!X`0eX6`gE|~V$`5pT;CbF-u9xuNK1Y^C@xI*lqM+D1D3S`)^@ZNqB`H9|JDcGAk zqxV+bInVOu#=ku_&Z?4fm{o+;jYjEl9*oo&uta*+brz4FANPKs^N3mCihDUc21k$B-am7eO*@)!XTG&JFa zG}~J=KYT)Cb|qkEnOOLaz9OV){7J*&MH*|NNMk@6Ir`_r*(ipDy><}dUHooT%A&o{ z#^GxHrI6=(7UTo}A1oY@6HvsXdyFdVEnSzxPg!_MWufxXv6_*fn8Hs9WaApWP0#Qo z+aG?GYo5Q&NnfDE-r}8i;t{{_iHJ&E#hWYzp49t?4jP;p`PJFA1YOSZVbc}tOq406 zv-pvRu4!83gy&R*)fa2iJQspk4*k!T>l{=&sYBxEy_c@!tT+g#JePIKqX|XWuh#wp z#*^2rYZ@T-$K4}TL0czt*m7l3tZk-|^omQ1BqG2hisCRhJwG*YHIfxRbKm~@03t$> zIah7GQ+U@sisovru7->4vXV6CS+aN}QuW22`iq!IrJJ&?{f1!_%yM11w~+gigAF7* z-DfMZPrp1P2QBaU$DO{9OIB{ss3hg^^9>s#>9QWZ`Fp}so^=_tJ1bM0E3Xzv=!gCMKq!^jAU` zON*l6f?6Z*SGX`P5-_!>BJHgjj>|$<_AeJwr+#L*Sgf&DqJaNWo*AH9|Lqd7d-xO^ zOXIfy9Sn_jY|b;bVe#Sa#lj-n`v?Xx7zs$$?1lDG+L4KzEH89zY>`(RB z{P38&f$$Fiqg&Tr1;-G;`gNN(2Jc>+J;gRWZ1TEzwRdh!O8dC|ihm9m$E({@t@_=f zH6e_q1yK6>F=c)%6P5Mh`VJFF2=7&wAkce^F^};0#nJTH{8Y{?5+3sw>vv+`O9_55 zu4@EkqY&;|Xu7pQlu3-bypQ@vDCNTd10pA9{H@fNSc^17ke~c_7&7d_%rAp)8Xhuw zDSdg^a>G6Z3^Ixjg_So`+DYHwa(`8yp+bVj`Yl8U1F^xH{<;}h`Tl%37a#_fqpB9J z@Wkx4nddPk8q=-Of85Xu9{p#*) zdEJ{Ky0itCqc>*_Z?@)m52ZQ8eZO&#KzEbDTIfF;~#+~k1Y5Sw(Xr^8O`!EsqQ%5ZrS5rgq zr`i&sBX{HNw|8hHX@19qI}bJc?$?zd-!niL70l3pb0vJf5rj_K=l|?>wLCvrvTwR3 z5g|c|2ef7DW+2@ob#wmI!MM?zvu}CRh#Cgw&3Ghvym`N2SXkRv(_wqQzcq8EH6Mch zCIO)5m>m6*IaTw5X_)w?Q9cYzlDe319YF6B)vkCP0J>?U1_SS>h*G-^yr5_6CJrV%4BROt&2)Ta628t*4- zXN(+N5qs9K1&FHLj`yDO=8IK^pG&p(e0_&&$Msrxf@j2(&zhM70?_@fX53+$@$#x# zijJ6h{JUj_5jljxxhkw<5tyvH$XXI{ug5|G^b+y9eJlJ+=Y4YoDm_TUS3VVMPvOY-B%kMzz*$KD!eS%7&tePZ zPVCNNjqo>OkizM2>mP!unTLxpi3x(_?-2f5=f4G7M)buyRa~~S#qO69)ueDpJL6w2 zN-7rf@}zXW2gbSagBwcF4dwsJ@I-lQ>W=BlyL?eB(k zH_pz5KMJZ+q?tBO^dF*Pp18Q(p2DKbm^#;cn`cBM23z#^WK4&F6IIJU$c=fy?txZ< z>O}(Ge~PX92KIsa*eg{noJL<-yZ}`nd&HSAq7J9jwP=lWc=2YME_Mld> zRkQOLX+Z#xIDa5z#0DY)t8yuDDI#2PCA(uz4RGwHZ~Un;fW5KaSDvczCKY!Jfo_Rg zd`!6pp{&rypEbmyrtV@bdT5NOaJ^(rQ0mi5F^STw?-W)TuLu;R6rruT$DqVY44G>R z$n;0`_BV*!{!dO9>o0*ZXFQ)CY-Ld?n&3p}BwK~p#ECbR%5Bbr!#ROY^JT5L8yqx1 zT@p|y8x@&5OIh*4W%Q-}f3LV{MC<*b)%7@VKFOm$=3!S6eUl0@LaInN2tYTrnk32geN}eIDhAGKP8m&>Mn2&*rJJQ`gi^9_&rua^m5z9%RWjRBM-ry zfu>oB3F9hVTf?6Sr^h%z@Rg;YB=Ts`2?u#SmwZKn60qPJby)pERNd6#ikM3TF6jPC zL349Jy->Pl!Y@c&XdsFD?tErH1mqPT9N(KRuC&!qwVQswSzEaziDrRl<6lUOG|}(( zRED^aqBq69FRj-AYsazW4%og4cmU~a`KwQ1V0lOH%&A#j`bI;}VeSB*TJ^Xt(T0!+ zIOdHh1rC{S0BZB|(NOkW14w`=F>-}zR8 zq%V2o#_gS;T>EXhBxIpWu>(>;<9aAu2Oa3OJZ%q>0O}riF?yMg#9uHdWPk?!!yo?KkBm znfD;fAuE+mLz`>+BSmO-7`wqbx7iQ4s(=%U4wM2pyo`W|LunkSY(^b9(Wo(<>H^Gb znZ{g2YOe9eOR1L9{dbnSeEg_253hyT9xlRoRq@68Ar!Q`Tg{exsvDK2)D+$#%}A<)cR z5Jl4WZvJt$%w9Y_T9QH=r9hMLir!)~&?7Dh8kar9()YTz4zHG)hP(>AE7_#?g2a49 z7AObXa&eSyuPvEKQ;x>^y+=;Z!Qa$;$O4cK57&)5Tw5H^Fn4nAR*otQCCnvrM+Qqz zPwe%V_XzNFC%>~Cv707JOXc&Mn;tAV^jzprat;^yv8JBaxgRXhvaGL?*0Q-CW3q(H zWtG<=dXp4bJ`DT!E;g-#QEHSig@>N5EAR66^Vkb)FG#i7`CT)nmYy@K>w+{I-# zoyy{v`MSVa&5*Z)P|hl$K)))t+-1-ulO{rodj$u|^()_fw8q#JSpC4H`?hTg0Jg=V z1zQ}pSfLi018G>J9uh!Q100+emeY;Z!la+$AU+T``8*@qn2fLPAF&~{q~!aMw^u#R zL%9|Cx(~TJCL(LuSL@c*b;IeID)YJE+jV{Ehz|A+W+~pEKJ4Gb@|7(%&VNP6tJ#!^ z3;*KAu(fJ$G{VsE!T=xY8DQt6w~^0Pv`~8l&azm77nN>xtifhmyIELvu-{ss&Wd@y zWtL~`M(M#S>;~a>JrkwS*&pd|Jt=-N+JUzsUA)QH&U;Hj9^16Z>!rQt-c^JN?heB2 zdhDaQc`PMirEAj(?01ehYZi6BS%9E+OZk&?$x9rs-;6XpJ(8qM;Rn`2EjKL!^2)%5 z@1n9lE*Yocju=HM%31#9)zO6G%waDHy*8K^v0T|tWjtJKIazE!pTjU~IEL)u1(b17 zr7V9_h1j6S*qgY|avwA9LRP;+RZW+g)Ek?8s;08pe@iZOtH0VTaCH&`}&8e3_2|P$buS z9*j9a#3l2GOMAAN7UI!o;XKqj^2dj?MH(=()LQjlwT8*}QB{f1<-~+frq`7%B9dHk zmTyM118N8J0kuppg%arc52LQPg+GiM7lYPFg^$^Sw`7|{7Sa{8X!R)B>6dL)Jx4-G z%7N2v=xH8jlTR2zI3WOI#y$i)BK1HI z@`vBkp6%HMCQ>jE2MynnAQxL4N|mR`@F{qUlz#JBz9RGJybo{h?Z>W$zI4$Ml)YMgX+h$^rtr zU43zQ%{Q78_I*IFzgPz3ldgPzC- zwd3*v`u}k@m+|i#EXI5L^S~(zZ6f1?b5ZN-AxLPm5QF;x!-*%Kd^dwXZ)(5n*B{$2 zq4p__A`&{;UI`rT?ss*luyN2BbpNTEm;kDL0Cur5N7P8b_uu2o5QFCcOU?}-)aP>C zjL@zJ8<0=iv*o9@p`#1WX8&UqV)|e42ec~9HGN3)N9|l&qyj67eSZ?zEh`lORR0)- z8o4yJ3tA@SV14JPd}e5>o0jca_0W&^k)Z zH*m&pK+w2IPe?*s@ge}N=ZDB<4jGg?NrEQtPBmvr|LuSZIL78hQ;cF z+m{Q?3gSAh7TKG=9~s&fIe~WtFkZhGpxC~b`UZ&xjmr}~K`^r-LG@de@(P`*nqF5z z;IvzGE)@IvZv69QKW4;n`ccKoVzbJQ$o@`HaVVy1n-antP`YXyF~LZsO|x(LJjYDh zR}IGBP2pD3ku%S4PbGD@=E`Gw3lxG&Ce^M>2qZ9u3zo{^*HQ(i;5B>Px|**JTh`8|=6$x0h;%8^HOv%#03nG% z@p$f^g2&!BtM^#kYEDhX();^Zlvqw9lSfwC599*RGNd&x#tA%JLieevi`O{<-u&U5?tRist3*!A zvMCeVpG|P+qW_r4unHas4mFVton(N)eFY&@pc`($C3bUiw2A)2%;m7gbQ7ByaO3fW z-B>&ya%5akk@;}x`J=w|bc*aK3;D4e)r$s17k__X!~)9z;xkmjuBNoKqV3-Gy5Bv) zVG=RyO$`n}cAmbsFXgPEq{EK#NdRwm($XzEE(REA`T8O}$seop(MSV2gxni=>lA2@-vpuZ#>(s&1e`>X2d@^!EvqGi30#p$&w;Zzm=DPvqO|Ub4Hkh(fr36tx}$3_U{#eg**!_!<$UHxZPRwy+NpFkAte zr~fOeULk=OXWT&kFL{fGAuHm!nHP@jo(Hkik^Tymo(vv%syw4c>?4oEFC$e*rM{Kp z?@)`_Zj!ZOJ5m_fr@27_XeHULQg1E(}M za3Qy?Q@*wpLFsD8tS^&e-OR>BMJOy>oPqS3QR^v-ru>%$Wx|1T!j2DusfM}V_@kz0 z4T0BLynG6F)AsFjePLBQX49oqFI^WH;`lqG_g(36DpQIE(1yH~ZsD1t{`>Lm}pf#O4S&YV{XoK0Pn~1^$!b`jkJ8L*cp1!y^}}yeIWv*6sSI-bRk{ zz{>I*b1D1tUN8S)Dl5oHAd)AmJRdbYHkYU5QGF;vTYc!$qt+__Y6A|XnT3BBr1XQw zbc&fV>r+F^9a4*)z}Ho*BV9!Qv=;<5$U^u??gZs@eNG2zIEn6|fhD?l56ISVc%{9j z#b!g3r^(FR*u8%{wKT5tE^Hr|(>sIQF8-*VS-2&_Kl;_ON(M(sX_8xz@PQ4jc5i~V z9nD)%NatrnPRQ4$F-HQr>p?(X!iod*YTFkuyr?FTqt&kGOPr-|zt<|k-wtxCSZcLS z18X)2^#g~&lWBiBHe(PU)r4y=7cc=9E~tGmA+kD9Bx|Ym7Zf(X?2Z2u-xm|$=pgDq z37*~@bu9s^QGs$xqEF;Fdq0O>2D#xR>A|YE`g_7uR{I& z$}`)yF#T-*R0aHc(|3J-Je$)1oT=>n7sgali1&+hflvc~aTPZnyuXJ0^kkA-xaR73 z!JOkiJ>NpFdvV)&J=XZ*0GDX^0%*>#u$wdHJ38m{uJ(_k+ZkD7>;VOL`|JHD`){gi z9|z5fch1<6{57%w$H=`-r`Rh|4)RPzAy~3u7Z9$IZSyH~t@BT9Y`rS-kIrgV&?2I{ zWYGU|v)}Z(dbvb#DcW#RQ5aCgPSz?!i5r(k9F?*^EMujBeyHCZeFBrvu_1Bo{Q;Jt u<_SlZCv#FdZZz`v0b2k6pBD>V(Kj~Os3`jojb6WhN={1oX}+W(@_zs}-a=pi literal 19005 zcmeIa1yEc|*C>h%F!&H$Ge80a5-h=WNPs}l;BLX)-G&K}paFsg*WeN?Snv>n1Sf;L zy9K?Q?H~c z2!euwE)GTqYI0-j8c%uoacQ3Qc98pX^E#SAEGfLfsHw)*yE zCMXy!Xl5p829z~|(b0m@K#4RoG&K9&*-Q`xfB>4!1fy>e+<{y-vc zfKEVxyt$cbY@BLoxv4R`K}O%Axvc|`0p$%WWBDGy9&Cr_TFxja6e+hCs*Gz;9102) zMdley)#Jla#z~&_=-ZqLq@}6xKqNzh`SPn+i*fddF{hlbRP|C$41IP{ z7wX41{#|7DA#)|*0F*Zj>8n9{R#^n>t8J^*OL>5R zf{c9Wu$c+7;cUiFnDqRxiG$Z&8kl5= z&3rAl10(6yIHzJ&e9gj*3_ny@IsLx+m`gLp{JC7)Lcw*=s9!Y%QCD1+60xfOS;873 zL4pX@S259qg>u{j2BL?d^3AcDWvILL$DSUjHD*M|;H!BQ9MYg%@Om<`=sxWEi;0!p zok1!D6r;7P+7AS)H@pwtD6jGC3V=?;c(K{qw7`{i#?;)mKWo>YxCS{^3!;J~IN{Ot z&R2~y3)*#e5MAY!WFXi6L17bdh)vIukYl0&Op_@#(8G6bQENFrYtiZ<*s@;qVb(^a zYSV3R3wX3ozt>2JG&1V+Kt}7bj78A%a&^f-vZ}eILH_tmLeN-XIxj|0aV)NozIceE z^nfeKr+~anJu=R?U+Fu=7)K0g!2v_;QVEH%2D-Rt+J%?OwT!3#9lqgMTp1Ui>g|G4 z{qE8;LCcGF<|N_;Ld!~~gr$38huuj{ri;*wZ7X_k4+__{CI!d)PbAumcS@mO&#uBE z&dZ7$9=%WBdX{-3+VoOB!71s6znehN?uF?+@bbdwx6EX@s(8vEJjNG1A934uFasJT zWlJ>V)}mgq?;9w@s_wrpt1Ke49~ikF--`M0qB?7D5_m3doGowaBM%~0Ssx> z7Wl@yDE2(3nDyG87!ipX@F`PW=1T!<@=14bIupCR{})pMX&(~*P?g%cXFd5&%vzLE zJIz1+kQEgTQq8Rj7-e!%rxZe&D z5mvg+TD{O;8Lvw1dkytrVr&0JSdq-moceev%}&OY2`sz6vCQ)PX-=bX#q}9O+NB$I zTPbEhab}EUlfR-^5g+y2mRHkqAef7|)`7K?jDqjW@~f9MTMkn`Jo$SI^G)W)B{cQN zF_AKy@N(Ojl+_`kUN&(@p(~l|=JJLvoN7rSrgZtQfEOmJwVoRK68TU}Z2VtH@dMgXL z*X3IIr}K=imU-(5^&b=_^u?MV3R;`6lEal|q3^#_s^=0dxNY5hH2HSRp<2zEmnZPS zg&3OpinGhLq=&Mh&?owRFi0`Wte2vof~i-6?$;GtsoI6= zFqQ)?^)U$RI@K5?#v#l;DmQ-gc&`w8@220lpa@mGvHA{dLg2AIDS{W+kzS5I^pSFv z5ST?{@-Qp-I0m!Ql;?` zxU}|h$Y?cK!L9{TZHZFlnsU-HA@c>~r!K=%N^DvB3Tyd~#-Hrq>KFzvhOBhs_XY(w zYQL+HoAbHuZ85Ry;b|=YnH+c9#ok935UX{ZJ1wTZ)%QmgU=x1F1DzqV`#(?VZ4L}L zmrv*BCRfgd3!lw$UuFJ^eqk()E) zC_pfQWm;HvVUg1wA3u5Ym={&D0VLQDGbTUfS&<{uO0Rx}IPF_4v%FJqdQWNc!05PN zI;|!fHKWlX)>r+fj2r8Q93_<=LmZh)48C!)1`;DveTbn;>OMd0d0f>L{es%@X_(XI zn}IZQ|Ev5XSsp!sSRaAX`)Uo)=jri;x4KHXSIcjVW4Mv<+EkhD!3t*7b%pR9J4+O$ z#``eFz9P}y!TN%Ul^C>Ta2)@9EN)+s5GYf7=wP?Y0*?i}t@U0oOO9%&`+}hAVrP?l zKif$83R*~L-qZK%l0f2qYMZR1vvZ*v?78!sTDhLzb17IVzT7`oj^RYIpwrKl|a@Lo}2LhrEonly2aKZ#7d1gVE|tuGUrD85W?gARvpm6^2Q+ zX@@q47v~9LcFWK-@w+aiju0&RAgFi*@=(Zrs8g>`h-yE1=0}Pt$#etKcQKqFob(Ia zW>6Te^C4O*H|mhqUd&>etRvnuKLvcH*fFc?-X~`28j?)Mn@S);r+W%u)YHcy$UJ9%U=TG zBJ=r(+jr@u0mM$R0YpFhZMCvEdf}l21R+jY(YQ}C$E$_2Msxk*e8E7DGPv6aA@>aO z$k3Et)7o(purr%bWV4JpcWlX5v2GHLHB=rd`I2jcea%<)9i zkWV4H^X*q)iGkwjd%w{M77w>~*ePZ+5{5jc2S@HL_<=Ib$LwO(M!rVq(W?J&tV}Uv z!=sC|e+K!GLHvOX3Kpqyi+z?aWw&JovDahY6?MUZ&!IHUeL{h)B~_=~6XbC+(FoGS zJD3ikBXE&~cx6-n*Ir6s_)NQeFcj=-NzmuMt2-eD@iNV! z-M+wIJGAh@;kTqz0?5C{t1IPxZOaDU?{ydXKo|-=%H(35&8;hsf*yVUDgTEJuN&RE zx`nbJAa5#@gXxB3=!BT++kdrp9}i0&%>2&wo z43}`!&5FvKNM|o1#K~V0z6lJc3{KI6 z`~z%^@rF1yIDcg%hUQZd35wu@q?%ivysaUOHP8QC;Q~ z!mflt-cGxkJ4SrZ zgaw;W4s6{51W_q_r^mP(riFh!!Gu{lHM^2ZP2RPsm2}kPwozl zy_Q;-E=lkW7fWoH485;>BWH>BCeUyco98?>0;(y4km|3#CrDaeQS`1O($wWh&b}%M zA1v|4>D4{EVnPDH9g;ttQcaI03G=a{{5_Vo1LC~y+nifY*C@Exezi!0J9uxjU!lP> z1_#MEtoEEG@jnF(!w-fdk>bzos2}=@^g?Q&78&oV{@THT30H40!0^G_k-tRGHekcG ziV`a`ezbQoX21p;4E3gFQ@2>M`AZK))5I+jg6MNIi94*6QdmiLrtf3UfabYHMf#9* zZlFwoKe}aps2kf04L-=PVnIPuV8t0gc5ft7=>;Yh81DAB#={@G+n?1w>u4M!mb{!H zs0OJe$icvV|5QCnpBWc>*H>f&4t>J>TXRx57{?fF%o{Xy7ZkKQyo`Q%(0}ejjZ(z& zd2{(uI#|N{k1|4KX7bF15flm$47CuJU|^f>7K>d@r<}#bJ>Lx=IV_eULDZ_*IHkRY zdYH|?mGG}NvT?Y1DfDsgVEVnMh7Z03M1X3+68A#xBc`8kH*BYeCHCh zQBrGd80##u`FUFVPEq70leN&Ly-_+jzLioM>m{oKof@98>`ae>^7Yx1b1X%|YQ#OT z`kl;o&X3-_hgwV=-v^`H7^M^AtGqsl3^G9f0Y{ka79S13xO-)v0#?v++RdNK;=4sD z+_3KTbf@DbaW2{BXkSkYuT`su^z;+&)f1QU2pQh1Pj<+_h61)u}} zV<}N<$`wh}@T{U(FRA84UJLpfESn{c8p4$}uU&1N2xK7}YD48ks8~4@N_5mu>=X#} z@p+pJPOOsEe23Fw^(P1UY|4AIvo*A_;lX+aaLzAeCq2sQLGH`1gAA;lINcnPM7#H3 zdm(ExRf_3;o>U?yBw%oxC>TCxi+j<*LWtk|@e$2d7?q8cIV;`jU11CfJ;?8!>o_~! z`#ih1UgkHWBKB%t`@VDmK%=N^s1)=R_D=LM|1VpSY92s}m}qh+C+x+%Z?$V$;N$dR zG!(ukkI$^t@GndZs$)bi6nSp6-v?FT7D*q1TVF93DyuJIOXv|6lfeGAj~YR@>_Mq6 z>kib_k(zo6*wQh`PGjVae>EpRg$Zq$)W%mV#=mqaAdR#BtR_LJLIhpxMcAXK+|NQz z=^kaRX*vfuTs^TzV3oO68sq=14h#bxZC&)2ew47Jbliedsd#de~|E zO`EzoSh5?rGNgeIrv3JJA_tmrCAw9O3_>s|d3KAByKb-(-nujzxJ(z)s6wWSDJ))1 zM=P7py9kyB>qXq7uN;6Y5e3=Ho>H*kVwoh2MvBf%_H&Zpujz({7hX~wcEdN0#ncP* zHEeE3RGO|;Forh;f9;<8>GMY%@bN~Q{_^&Cvbb{NS>RWS&T^= z#Tu@ssk5XNY~a`URU^T5_FSh~VA8Xwl4{^1!zVYs`wPg@u=Tv1m)l%q+Cy z16MtJYr5V8cOg4IYb|mwkBS3_t%Eci2~&r1@9x%#*VXwR@$>!E!y($AF*4~xU-Z>U zvduMsj7-Ke8-#Rgp)PvlJri2jD>WOe{)etrC>DEGE_-uom-pHk1a->^=_POI7ANj+ z%j|zzwC9kUJ#R?mW@)+0w)o&_&PHG#hldd|r+v*}yr)G6vLy{R0k@5PtiQXMqR^ON zq%!OO$)u3eP1$?6S4bmlt+nWhm#AHL5>?pS<5i!}UH5eZ57;poT--zN@o(+Uh_KKP_We*^WI@q#$OY654~uJeXlF$Ck|1@_eraqkgpzvDd1R0bA;Fz* zD#2@;Pv{!)Nqw>5QaA35L{tLfF`BNr_zhGt59_p_oyK&hiKTgJ=@4i_WS<7rR_7<@ zi)egh@no)VEx%G$^1nZ2ww#wfW~(!~-ER>IFRvHu6$(6=HfNL|MO4y^^z$bMQgp;M z5!=Jg*y(t!T9$@6+4yK{?T``E+TEjH|902gL(alK1bGk8Bo#rjpe_$-zDp5c`oB-{ zWVX!$VD(b|j#USisI=7Ic46lXh}|F36sSfjdDE5A1L4k2 z0tpd9?{^FYGy;jhq_%$Nk#6=4@eu+oc~C3TvVjptVtbD_aTHCJ59$isXb z7j#}U6+;VN4kFMsdky`hYRK8JNE{4R?Z(`8CpF8#OgSH0c6kv~t5Hu6egyb&J7*v$ z*#G0V*6-v5B!xnxDzLp86(X3$$1jF%-R3T^2_CgG&l#sdCh!cM9hDr!YPuy-z8{7y z(IiXd=DP1)Yx&Ggj@0kmvsvZ!hZy(x{jp_DoJNsrVG-Ei+WTAAYf}-@IBQE4@{`-x z(qWI-s{Es^Uh|>74y?{|##1!QRvq6wrhe@`zHd7wh`#e?I*+TM0A^aaw=*4H!?=^N zm8v!)N@vI2p~aJ?Z8f8~S<@%j>?ZY?6m;XBf|)POU$7;1jx7LTovHVu+BM$wb1bq| z=k>}UJ|&ad8vHu>S9E3=&>-J_knol!j*t1Vni8-B$n9}uJk^hRK+H{dV1qS9ttbG# z9A1?PF15r4z)(ag9j1`N2S5A?7&*U8=VcbKGscq>_{ol?y+eQh^Noxj;HM|4C8D-=5$3!ZJm)JK1W&O zV&(Gh>3$H&{#0Obc48j(*H#_J+O6GieCc2>46|@>Fzs&D+2PuGHODyeXD5&2YTyQ4 zsLV)|%HXpGp)X~G{TNVn8DiV?%^thj1ju;8G;aMsBD^xSfXCPGumNU-Z6Kdz2f^2ebyVH#M60-PE&#Kc~9Z zO4}BHLd1H*25*JZ_1D?KGIPiZ2@SSH%7x$viok^Y-CfNof2lt%=HPDGye@E@}ClBB5_#Tg*w3sQ1ZHnLx>p(V+ z?bOgQvGPWms{DQ(NKe~S&jQH*Ug!8pYyC2OgL5GSbcF4y)My^6hQ9QLrc-D))?AOH zw&lEs+Mt@V=_dRiXL~Abh1=dk{H;CAVo$mdbCeeA7(3lH%{hCo|L{_Nk5GSV)IY9u z@A~mS!lra&-Hx^3iz5;Hm})e`$ieXayd2T4t`ye(W=kyd9De^0()Xsmajgyp`S~Z^ zf&Vth35}BeiNJzhHdT&0sS>((htRsBFBvY=W7KRw99I6srdWom4sO%EwiWsks`n+1cZS4GS|}|H@ua75jTq(TZLlt%p>t1%^~{ zeZt86fqI93G&Xj2Or79Eyche@OwYYJZjr~mN|2*nF-CCTpCk@u&A?-eEJpC7lG#E` zv8u?m(6`Xyp-%MncAn$aNU((SpJ*d+`w%bL1zSRAxX5ZxC4mZx6(5(6_sc0}o(hP3 z_%J!&k^W(eX~P3-y~AdKAkqu0@Ce~vDZI3S@Z60 z-V0f2vt6gBkaMQL@=b5`S9Kt6)vj82GuI>h{Ne;tSzfD)y&m&(Cr5*O^n5`V1OH5? z^<4OTs@r?+@h8x{UOrX1%iQZg!soER

|sBPd~Tq)Uy_g5RI z2^$+103H4(PN>$T6&@2J8wi$yd6<`Bt_8lp*)7fjeCm=LF<<^+-~4n>_2B8U*Pm_q z+-(0&1z2%S*>=~BL>TX#q+s%h6Erf;)N7NJ`T576&MUu}s$fZ=mr!krjoQv;MZ3{zV!TAv2Q?7Rk z`}1bXK#X||9Uy3jM4*&iO|6PQK>&}EqTGQml1n9sRt00pPzi!a`#V0kZF2I_ekmsd zX)Y^5Re8x8^MT-9o>i$akA&j!>Gi9C&nq`0ieB%&Gd{RE0l3UbMN9;umii7PNf{G1U;lj; z!2V(OfFHkz$7MUC{lKThLfG0PURo(~#Qec7ChYn1nd=XYown4n<F)nK`a?|hR}s;3_KMlkjTo1`gpiFaR^m6IXAz7iieM-q;#=PZLKqd zdtdpo_!8fPqiiaseb{v!%L#QAz+)7OGD zF3l95R3Lb`IPB|;;LXD$e6gZe-9m6qq8a9aW`nMQ$268Zx{(ItzDH(4`i<^<8^Yxc zPazLq;M=V+f*DZXFhIe>bDnWtkG=L5N>{4n6MnQ_ab&l5?z=fS`2KV?H_nGnBv5~t zdj{d{-5x-Lb7I$|3hCchnE75oi?%W4CMW&2NKS;k;e>q)0s<5O+qpdDLCAa!{XPVi z2#`oJgdYNa{bjAXpf2v$!nW2CAGi(lHExbo$=*jx^WDmH_%n6Na0RQAZZ-5zuR`41p0pN zIgUdDFUNXyr(Bktu=I!ksh7J+fIV(b7WQ^!8KlTR^||pnYR3R@*c^r)JZ%=tsSSqA zCPlwnu)NS1=}vP72Yzv20B_p6M(#>OP6B@`=rz%)qj|o1=;ij)_6V63Nl+9&5vb|o zO2B|}Ade1F-RUR;5j@IVv|sZ+0UQ$+z^!4fd=O`HnHPEVfYkO#`YnVNjHdeQjC*IS zHa{da>b@$J;N&g{^#hD%OAL(0AH+Z5KQ(1~wk*98832_FppQ}tf>HEmt3AO$#cb^l zSwuuRmacSu*yZbbYM)aORHQ9b`S`~Io-NX-)X1%gvxN(T=_fimTmN7U?FyCIIPM{} zieZ)rXZimJASfqLaO?$UfC#v4sTtLf5YAa|k1`PimD{04If2-tK%wCO{uK@d0~x}# zwaHSFA|d+?8KkIAxnD(SmUmgxiX-KH8lP3Js2-;)m-%9&dxsqScpP9Y z#b2LAI!76-`FJdm=G5_2ZWk*RFF~yubSleM?|8qHE7%-wAqY}O8XnrEUDA`^rDX1P&BKDV?h%-oX-pteef;YnBfu~wKpdmx=7D~Ne!lgsG*29K=u zRhSU-F-KUNixjvv-xIyI7BD$hLp1|6ouT7qqAh!g(lvP$VUYt~a!Ec`a!s!|n;@>r&Xh%J$a^Zfs^9W3ky7Yb6 zE;;XA5>8wE=HYgEue*6uxM+hoq}YRww%#KC0S-Jn&+=8CN^qh_=a@TP!xpz`fs@wM zz5$0fsL9wIt`e*-KP&^J6(y-=!yGG8A1ravzK!(pCOftu34zaz$S8m^4TfE~FKxSB z1YV8XerTM&<&<}H=r!md5jvi~e@)GNw2fd-U6O#%@X9l`+RFVT*RtF`Ni`opf&v(g|9!4J4se2Ue=1*!+EM3; z;$1rXr54wrCqi{1)7o&(I_*&Z?4aGTRjgU9BpJ9-$ZD^sK}%gj-_Fcq=fl?Sej1E< zZo!$Q8(S3E1oISFf(q|59O~)kW@IILHjxwrRTUD2bMn|JR+UaFLflrA&U}in2=#3b zy;=s<5N$qt6+zGuamYy|48zDb4eRTPO82Eg461K$DWWv7TF8|doFe*S9?8!K} zUilEw`f7kBt{-r{`%XJ4kz4@98bImvM_fvy8u@KE{HDs*03?+qhGm zyNIffT z9SWDlKxvBL8GcV6$x1>b2ojTUB}7ZZusnZ(U|Oo*g99feM*BE8=puuy-iO}if5#FK89kK*++eg;a%zw$WWsStsVsb!T}SAqx<1;A~Q58Aga=Tn5| zP3ggJ${s7;SiT#OV1sA;{A#B>>&;_ByQ*-^{&e)w(3{7wGcOglrXiCyYuWReJaAi|rIq$1X&@n2bPc|G@&z6%osgkeOr% za1z1CbTx@t_+AKQAylOme&;`Msee4TE+3V^%RN5h^LxTYatN+jS|5F}3 zuMe-GNw{{oI@&q04~e?@aP90%_kf0PB4C{p=^+kz8DF8X+xXzR+>f>Yn&GPb2|5Z@ zlC3Zlj9r`mMyI9p8n%D(3>_so(f+nBfwVThDHYW9pBg05LTZafS7^~hk91j$oqx-O z-}x2;e!$rcC;AFCAy`6A7%JTSuqzZe2b;=jS4y@%%S?q*#eA|!zDe$$4H%>ATFP#8 zI3)H=^u%jrzW2G~$;RIPX5TAx!;=obs}!5pbNq+9;ftuSP%WnH_HBQ{<`>sZtbP#0 z8_X|T4@xjzy+C~&i3Ku>jerc#299j^4&gYxw-c4oyZY71wba&Szql4{j*97s)5y(v9S?ZLHRsWY`qe%o?i6M{ov(C@F#NJ-_<3<8oKDmK?~+jN%q)J(q+T;#Ngoo00+>k)&RR@^p~loB$lG6uTOIG z)?4vJ9Dv&TV(PKyd4G}Gfsh3BJL!U<#7N`e{&KHk@+|2%^E+{<9L|DaS{#s-PV1++ zYxU?pX{mu!H>eE(PzGrH@oDX z&k`Mm9TPF(tNbfs~;!3jQ9906t?@g zsvNQe&=t1yz-?bXjdZYZsft4+n8*>w9Af%F%KCu?T{w~dW6=*9%sfCwBb)-O?_-mI zS)0PvQ+DwF_kl~p@V?{WOI$*Ick}9oQ;g>?)C?aP$pfyOHxEa0k~$ol=sGE z>xU1Y3cT+Mj=R35N{mkq*DI!V@$Ty=>u6Q*`s7&tm8xwx)&tEWvH8$FvpU4LCJ@4T z@-A9}<|r>8OG-)#{Egbbmc+S@)+XfbH?$!uAR4@C=Oq9wWN=*l^tXBYpuI0YD&Jk- zl;djj;(*tKu2Ht>X5aEjt7PXb98M@Y$CKLEGCz&t-=&a~y4LGbTqb^h48iOgxsiup z(OKBx#(rwYg@22f86Uq(*IApE-F`73Xkw8T$jOOWIu7Tj2#gPYENk;Z3u31)EUOXg z;$Dzs%41X2e}}J>fj}xZcb+D3a6b}Wnehy|#jGGKNfZ9!MY{lJ<|6Iuf6if%EMsjq zAjBz9n9~>tPSWUR9sse^xj!Yx9ggGvMZP|(E%jtnb{a___aoA%5^eK9tCI`J>^jx0 zOq4u9*X4f;WP`js@}by>-m^!->fQG4<~(j}eJJFS0J?Ml-Se3r370Go@WoqPX(Fr( zMZ#T{;;iiP9?(Hk`r%o(T`uKZhe}85%y7F9&gS7-e~KE(^tR}78*5R8jwL4lK8a)b z>Cs-~w{t-IPhMIYxze>rshiXe;W;d_E*Q?P&!@I>;v(E5&a7H`I6nxL&=X*0gt+^> z^n``TL^NW+(c4k6K5DpY6C2aWQG08c-e^S4@Cx}Mp=`vf3}E#m*CC$;!=Fcc3U9rR zYHCHcX3#PPDzK$WXdR+YT!%;&zv!4?$r0}66iC}|-fgi>RJ#$UNQ@ujl@)&IoB9cm z%9q~u!KaaQGnC;q;^5mBnJs7o49$oifa7mXSc4~?8E zyH5{(3NUN2BTz=i+c`X}-y)z^0^kjh$Wfp(e~@0Z{Z@noan=4Nzby(q`5VtAXuhN#G~hHkZyw7pwGoM*TM1DU`d(+l-v)gw zpM^VA4-z6=FC_?5XNf>YDBl?~2pl=NGrZJk) zKTLoaLojG^+?17!>=#j-ihT|4AbQvOTWrziS3FlY>$uhT8w3=9gV-&1KRI7Luz)#4 zr@m=9*Rg2jC6JYSNPy`EgJ@nfHxe4X-vpjNI>SpD35Sl_>xl+L9_U!oTHJnt!q2U9Vcek4L`9Nx^+kckp23z0jOzY5y4P&wimdAmI>^ZVClLr>jLVTwY<3lU*f@v{x35Gi z>51W&RQ9FH$u1c7f}ySi5O|(F?1L?yw`ZS?IlFk=XqhiPqb6RZ!jSN9pQ$`F&BQ?R zDL9vnVyU<2u{FF-;ppwn zd+K28-Q)UgO{vA}!yQe6==ibgyrha3)F!)P=!z#7CoZ7O7Hs?KDmQ~NWaRy1w9JF9 zv=u5IVuyU|2ilO=Rj27~{EnOs3G3-&p@G)s+s%HpwD?ur7A=`VsmgZq8-dg?3TA?2 z#)jOICRbg9t!${~lpS~Pw;V=%Ez4zNuGbQ5@!YJOSy5Yzb&?Z)yyvUKc%IOwvkh!(1trJ9 z&{(<+s2JA$6Ez{7)l(u60EMRu>8c8v`8H;gi9y`14pTD9;kS6R)y3pEXwp`fX&5t> z^JT1m_pzyi%S#34i1Sl?#_-BbgBk+ta(ie43(I2#nolRlW?R~WLf4z;R;@pOz31c0 zo!M%3jWcn3sc_@rK3LjkN&AW%#LJ~*@`0aoFeo#f_aL}d;FqUq z@PIXv$Y3y&TY<)7vla}0=T4&jqgAYf$icU?I%JxkEtU%7l8xqI6Orn{n!}&S+o#}S zSQp}s^LDYejEv5Y3Qh*f-jchMa=NY$NcV?DL&I<@^O#$R@Y>>JGt-xhhs|@Xt2%!gJY8V=%A7s~GyFx@ z@;sabXhwz~JXkMpHCi~{X%xDBwRIk#6c%Egh8`6~`cbfQtZ{xM(wy(G1&mBB`*4f5 zfZ?3PL_^_G+402@5vk&7dXt*J;q}nhiCkHS#X@Dkbiz3~0nfm6dvmk};8dOS_?+Wh z$t(j{YUZP%%VmubV60=<&T%EGUX#h;2_25@IR-%ab3)uz@uFsVZ~Y6W*D@_U8|PQ} z4UVLaoRBMjRPD8ST8#F^+K4KgtdBIW##?5+0HY@HHpp^#>ls$Cw}-UR2c$eY-}|7g zsUtlO{{4~ge{t;lz|iCG-*Tj+rZZXyX*`&~y8Kpz3tu|sr2wW+=+I-n34+zKp8^bX zoJ$IHoG%awMaR18{pxVTPr)ApI5UPh;wT|?4kYlo%-Nl6X!sLV5b*EFwF2cibs7#} zkq!MWYqb72AqyWJf#A&HIbg^=ME$&Py(K$c9$6rABLIZ!j$hO)UBi2`07iodBCMJ7 zM{vWXR93pbDuvtuhY0X3NPtHpY{@{iEjQL=q^EcOah>LB7OxlElsj6l0q5w!WV-`z{c^w#kTn9*k-V1g0uE=~EIdw8vm=97>| zmi_k00!v7k^dMz2dSY?>pYK&&0rDdPq+o zJUk^agJz60B(aOfFGY^#P z_-N~IjKD&W*1T=%Xo7a{hK%`P@aR2hKSHrLJU4m$2A{yiz@ z68WLXHz_5()~VfwXxN}wF9A+2wvz`0jq@!@zC%pl^dmdI>BdVgzhF^6r4z?-K4DBa z>~%z9PHzN-2tl$fe<&qZ;*+K3K^E{PId0SFvKn`vuGEjF2IbZB*lzK`2SxagaS%*Q zcd!6s)GinfJiA|d`?~-G%mh@R`SXuEbSJ%QbJqz!<`>Q+Ry#A;fqg2$p5iVMiH`M# zXTvmM+7Ecz?D(RjOFBf+Me6N3wV%4ypiA>|zOCCG2a^F72rGyVct(Na+l-7KNmfX* z;-{R8iBd946qg+<>b*6LwQd&2IeEzWV-?u*nSRFt0o$pi(t*sZ*%FuXYyF6hg3NF zxd)g9?rD{r+6y zq|(y7PUy%PxJk`sWpJ^_@-`%#EW~=o#fLdAIQjfI;#SUDnk%yAt$`ItS*)E5;p7Sp z5*XS_-OLA}L9iab65oY@Ax~)#ctC0e_>lA8EnZBx{50h~Q*nYy9K@dp5^0JRB_kj9 zjRFz{>>aJDYpbbeIUm|%FAUN;bhHF(%7NU%YyL#MrbcH){a(} zwMAMw|6mFrnQ*kP@GJ81Y5FRTBPTrsw1x|OcqvfEVdRiL?M#xOkqgxa&iFhd)X$xX zx|?mzWtj)T8qQ32XD9D;i#5)ieDOOc@@j28_4cH3$?ETu#>AJ{h;^VZMw)3rAU%OZ z*Bke-*=zMhd@)el2t9Cro_XenP;}LN0r_44Zq!d04FZu$h(!Y0^ojVk*6RRy4Za+I z&kz1lP6-_BCq;u%Oapb|kS;!&XY>%n4}2_TFv!TK9Y*q36Ic&2)^b@;@!2NT9wG|P zf&}8CIYP3G9(TPhhbCL?;tj_WCdy|oTkT6qOXLDWdRS^bqReHc-|aw8fEFIa1u_C|@N*X` z&ZyJsK6)~xeGON5V5c~&TAQD~?KfF}OJ=New=g?qqAn8kDGi7M3php}W^RU7ODa|J z(?BX+%-~+TpnSY?$QUz8;GnYE?YxWiXSJm7S2)t52lInD38b`DpMViXk|e&(Ix&wc zUCEjXCN0V=mXk|jbxSlfXy0#6yM8B58RK#Y4H{f9Bm~Vb*^!8h$kgB7WFONz8q4DG zr-57TBg}RJ!DcWdve*q*nENBnIsQ0)&ulEjncBFv5)aoDybHntJ29;?I z9zx!nmaWTWlR~fqX5m;)4lk#br5w>2Z%z2QSmG+IcY#IQZ~_dvTh@3RN%;qFr6zl~>kpfQM?c!yc*H)NMtN`Mvxa-8pnrj6_O>k= ztSHjcxJnR{&=I0d17ZiSGiFK;JSh?7du~%6A2;EfaX3*lI;ET+DQO)07nFlI4hk5I z_P@~uZ!>ifb|ZqKtk3{frtJedD$f7oH;D(3F^(OnAFV-QONWOng6DRmzFg`(hR=WZ zfKvfSy;$b^Q~Fj;9$vfbEyw+h3uq}?s~S(~JGMM>z5KOyyy8VoNlXA@NpuFBDI8g7 z@c?0S*v(x+LVLy7f!chPf12KR=A;<9c)7M4JV>a&N zGm*V=SZy(#@a5A+0gSwX+vY*@q*9L7?@2De%%*SemLsoseGeE>t4zrN;ug_;jD$?W zFG2V#5P;p7J;dcdn#N5M(7Q|tfad?cCj6hpv3G;YkBHGX3tEf6{TV13Nrh*n5{B^q E0XBXgU;qFB diff --git a/docs/README-unnamed-chunk-8-1.png b/docs/README-unnamed-chunk-8-1.png index f6691d4b616ac6402dafc621d4deb53d3eb22cc4..5d2402c896cf7ca484433d11a07ce44cfb15cd43 100644 GIT binary patch literal 17097 zcmdtKb#xp(v?u78nK8!16vxaAF~$%xGdpHx=s4z>neCXFneCXFu??M=nc3am&iBpR zcg~#I-M@Cuyr)vBO1hG+R8p0$r2DG~Wkneb)VHWGFfbUhpCwgcVBmV8Ck!$iG-MqF zA%|X$loiyapjQ|e0T>ud7y$to0jL`XW7!P#3&2?Z4fty;1uUU9P5>Gn2em*;XzYK; z{}AH-691Yw=wTTLb(<}rahjVgo1yM&oB&LmfIyrj)c!+^gPLaO5!VcLfpNg*I3U!0 z%~w`#wuEW66ljivTFbw5XpDc(=4N0s5NcnWUtin&{D#xhfpG%BI7?tN)Iu}(hx|7R z)Vu;;U)MW3Uz;sofpM?E=2vLY-*iy-wS2)ET5=c^r_VaBFfhcKf6q6v?m&0F%-8Qh8ClaOw?PWa?JB>zrJR-)!FhV z#Nhux&Tc)w9z4`;%%LyTcGQ4_8Bn!1fPoPcz$S+26ElW^hYNgzjSi0}#`fPNKp_uy ztEaM@b$RNhxm@I7!|>&<;<~(?qDz}?zk@Dp}U5L8GQ`DmBhdWK_Cs9ia*MWD-8TJO3yo}-qJh7yl4tGM9Fhrg*W)emSGk9;zW7c`}IroK!<>9shE7K|9=2EGeCUzZgU%G0) zQcP(NY@~QyahzW%HxTM5MHDH*wO$51-fd-a1*m$MCFugy1m(0BIq^Cu&=3NZa!wHf z?%kq>H`jm+T&Kxn`2m$=ESz5=!@G_ESy+)GkNzviC!&V9vSs~M?`qROcR;o>8q6ns z70a502?DDyP6yrwP8;fr87qJ5Ef4Sy=E-}UX{zrkU6{9b%d?x;6+0=4()C}lC5>KE z)LE8Z%TJ&tDv>h+TZNyC zjphdYNM9zuJ@_E|nAV&Xk1K%}bJgrK$oym@1A7amQzx112r*>)C+Iq4cxQoU$ zMXn{S_4p9z8Z^>>rfUh86u_7~wy#vR_1!F!q4t*$WTz!U_PfG=S1Y`Ci)ZA!d&{S5 zliJ=`b=u`Hd}vjZSA05XUX{?InlVN7KHj01>s-4*uIiNE($w+b1|QnwFo7M{V@QB> zmiCYN+KkDEP0oFXv6ML){-1%e<$ar2_60#3xKhU8;&_n`Zd{d(d-aHqP0jmpD0O)T zB>9RU^HqJ&;rq#L$4cwUCe^MX&dE;z@Ke9ZC%ie2grlZ15pP{%i>J*3G&>q`y@CHb zX*vZm`VlO=CTl|uSROC#$h)*{F6u(Mw4-DU_8Qf-BlZ0;UoT@2xsvfmoG_^@)du*0 z&on$Rfu9V%_x3;gm<=2eCEO%S25-3^Q@-xf(-ub47*IP!^}gVwso8Y*YFhYEv6YTv zw>nH2i`oV?k}M{<8rv4-Z}d8C529Oh%P#ZR(mIXt*Y>{$lvxLsVN{JNqkL1da1DI* zx?T*+X78_3+Lf#An)92>Cv9v`qZD7;9Cs(kz~{UB*l2-Muz(pD0qtU?@V#aAO}$kB zN`+LPZuOG70XbW*9C(Um4_euEq-OZL4J10;cg%m1)@a!bgoW8ZzHutIV|Kb1>mTkR zpT04?yZT_mugQbOw$UT-d{EQFfVLUsII`yJ&TDo;U&j>KrDVDqF+<2McZ$2R8CN=y zga$9-%|rB%?Ss+ADd(NhXXx!&&1C)OQFEivK>S3bM>R!3tDHHI_2V^}x-eriZr)$s z?`pf=X{4{EbMtZY5Vdx{+MrK2yl${qe{aS5;%fo+_;G-A%IFuQY4v(nzx-__(vNf$ zcg2Tvw6#*IJ6dvG+Vp>9LE^at8dxy|?eWOo3!D@A^Vv`iv1u*!Ki8cDggOVGZJ%LJ zor4Z>vFnz>_ZyZyU(&TYk|;hS5Vhx(FR zs}4H&VDRPQVR_;59-Z4XRu0U3U9c+*?q;uP~>;b8dwi;f6-Xr1m5m~cz8z;aYXxH!!JCgWM+ zVy+S708{6%uv?kpjY9jbWA$wesFzEARjuqMv&g(f;MbbW5{j zCKQ3td&P9YbYmujhS4hrvZsN8Z)g`y9$iC#F_tm@pBqD=*($FpQjH)d1ku=<8`@^0JxFpXKq*fX-%Oqno8S(quqc@|a)>S56vD#Q+y`N{qvGJ0D%qgS|=IH_;x zH@!NDLK~gO;<+}F*$IgTF;iK*^YjylkPm&SC`ua`-OKDmTq*U=q?t94^e3Bkj;XjE z0Cw4lnCT1uY?FN8@oCLj5V@M_*HaCUYBhTnaVoW5s`;lMKPtee`u)F?TW=cNFJ06$ z6;5?Q^>l)QG(XQ+Xjzzkda$q|{tiJ-Z`X^Ek1Xy&(@*8uU47$hb2k3_kLe*i{E(0U zwRORfAvz0NoZ=_Yds0KHYq|kJr`7>B0gS*%J0G2$G!%KQ)&VL3oWOBAKlUAg=wxA3 zA`)?WYNzx`)^8KbKLnc(uC31u7lq>*}F$08E_vp&&8(eoT(NE>*Ya$jRS$kBVX=6U|(pofGB z)9XPv29M`j8zDeiCKXb@SZ$9ZJ&*e2UcxXKOwAHp=AWNhieo zj)QAd?HlsgS={5yQz*LO;Y8|$`y^QbgOGkbs^6>2+duRRp@N{2df{P+9U5EZU}_zpZ?&KoU5c92y6cQc_0K_L@G8V7(q`jC<{&1$BLnh*l%?qC^}HFEwcWI1T!)7(5s}B#Xwh z$9DX#Z!5__D<3;3M8DUD&pQxHZ%;Hc{E*WbSUQ9EyK0>3Ba0to{nhpHq=i8H1)J=c zwa!-uk;s{XX~amv_>{c2!*ywU;<|pz5Y%6ckcPA@x-UVxOQPRP%=Gbt5bX8lL3TQL z2zLC439s0HQ>NOuR@UY-8R8Woeece+=X9><>&eSW!9E=2mEnhsRPYik8c&*_42n*UgQNPqJvy+XyF5%7hk`M&KmkpJe4l4pJ=s5 z<14o&Z5rdEZv@BMSK6^jh#L_Q#%t&Ir1UOtSRJlS4|Hq4du{@NG94aUA+4$-8m+`> zkRkf^ls<(=*rt&)r>Gupmx*t`r%z{rcW1MdjyW8FcW#Dv$FCw6WI;7#<}8-q#0H8F zfb<Iw98PL+3{@P-g|{~@cBnD+_-IyKYQ1wuR1=Aqa)cy zbdig4Xu5di^aOYKbh3POT0@|(kaH89wgUi-@+%RZYz+t6tEajTU;X<3P4;ARAJirZ z2E3T~BZPG!KHT1p3JxBdB9^QIe=d$K|KidtSyM4Cowb474i; zOmN8|F~eBxZre(#M<;R#86s^9?&HTj!LYQ_@zpaFEe}~Bd)ZC(_##J?#;oat$3eS2 zlF^dOA$YR%ZGqQaX!6)8XCMZH*x%Rpt2edhZE(Ip-@&y6Ad%tYYcf^wj8PM1zt6pB z3n2ZSi~5bjq1-SBUzQR)$*_A1QgaVGo|TNI_b=|mZY>7H3RJZly~~#`xiU;IU%$EI z9Ba6O%W)n?QDeoQ5fsDyhB3Q?xq?#D80zS846`oH47xI`YceE{AU+(NYKNK3zi}Gs zM#;~3f{{Q0kq=|$*?=}Sa9e3P(rd*d)eU(3=mbdXdRC7W(;JVX`i0O&x(JI%V7ZMW zT&eH8rhCKBF-FiTj&+((gj;(|BLD6~WsTLd-tZUttI1ASZAWVIs?(f2p|?!epD-I% ze@&K>W_5}A@Fxf?5%8~FWXf!cl&vD}GlVZG2Lxdu+Mrf9V?I?(PHTVRD9OZQdujVx zcuM&_<@rO=l2eqlx{vkKaRQzEY`*XE{&}kui*MMDC4ZXlgT1roH|rYHx|*Vb7yAnh zTMKu=iBysdYgg7@0tpTJQQ_$mj zrzUzw*lm!>Q_Hs_$%j<7yZ3g+>ZBPxrc>VbWOat2fA>OSBYl_LrAngrPJ8jbgjrp9b@J)9T07g&& zqKIwNmXXcFY1&&qHFrUw{a&KQ{@n(nEBnGyY2gJw<^}-;(Th(;aVk2ho9uA3C8c1F zNBQa_;ChOdy`(9H0VyWui+iPQMN3x)^Br%Up&KkOZ?F^qufkE=0r*iI(Ngn{;*qA> zzr=NHFA^wdo^MW6r&X-iv$^=jX4x^TROxqH!^NX*5>>rJ`nxl@p4ZjK)z`<9(xodn zCx-pr&~Tx^CURH4i6?{~K5Y$#X#K|04&Fa@&$++k1o=FJ#Ly$v;bIu|b-ZqwWi}DV zbG5dQgLCok1x(QWuU#@$-}7kjk`C)nChA!DUZAn8&{le`s2?0q1K6Faa38ut;x)ry7I$3VygMHSGaODv^cO z#dNu!UtklN?U~(dc7LE4m41yF%zF0uUVGW%Vd!kAW+H4WpgUZo&_XsflfZ{GBQe=E zU@Uu*u-cF+c*{(l^U_wzL?aThYh&xr@2Y+C)6SmOLL^r~;Gm&JVq#Fy&%0b|6 z&DUMRm1U5DW9ijU+mU`AUcHu@%i1aE^Ut#!&5NW`9{D)(Ib$%td`TdgzP-Rl$pzmP z-ab%}h?S=<*<;oa52|kfsde(nH7c2aam;T89&)ln2vvG4_8n&rz+_+qVV%O zT5cMmKGcc4Z+i!sIU)XOZ)J=c@@}K@kxVMGqyys3&IjlUY6B>>@9Yc33Z8csTTW7g z;pRU789EcceoCp2+}gUtD9qDTPs7AqHp3zOW#xqEJ1kLSQImh(T9vMm)n_idD8v-T zx{-ERS#yzd2Iu>7nh88V9r#e|8tD^7O3mMKpERY(bNiro{7XR%prwN~X55J_seKf5 z-nmm%t#-=Y+(a};eDa}!1gCv~BE&j#+0{$tK9C#|GdZr>vh9S7)hCyWfZKf45-A?# z=ri~>#J@;m{$`mO#}r+Y5;HToncw|N(PVV}0kvaXfZJ%x1O>B0dT;l`lSu6F>J zdKd1PfGF~mU>0d2W7pC9K<~-^iW?xTZ^-!B`)Bu-0%KC>*C z-V#i?PL(V%yS*9J;mlziICJ}{jny1#W=R?p*^QMoytOt<3rYB5mQkTS?~&MJ81dZi zSE^(W2z&b!DVXHW9v1d}qf~8w71am_BEWmM_~ZAFHo>BdWfQ3k;;ufh=+;q z%Wqpfj*R{;XJQPVTDW z&lU^Ima>p-z0N87qJ(p}21d#>7nLrGvmbE6@b_K9?zJtkpTS*-Y7j-*JA&3FoA&Vj zy=-a=XPdMP$Wt@M5}k2vJ-|twiY7+W3nR!KFxDPDP!%v^v~L(y(AC_-&w%wPKhLO8 z?fNvH@o?NUSyOvGrDBs6qhVUc-tw^@GO>F4NrKrTtDjSjtoR_8$b#g=VM^?QW4a{? zA004^`-JwDlUU(3zoxtBt$7`yl}9USz+sT-AMJu?(iJ}aSW?jqX_fur<^hJpLj;DW zeJPzpY%K|n_o)^aY2nr9>Ow}C4kOqc2@TCAZ~bahCp@*g+@xqy%$^d*DzyW)>Iea0Jx z0!3beYhzBrPFT;f14*tUhVc*O_z{tRbEs;IHf)^(lM~8FR@T4`z0;=jKTMD^BdMg{ zPunXZ76O_S=wvs=xSECb1Tc?=7UUT0rl>On0sq)@rCeU;SD#%yjb|;gL%VP&7wRBG zSK?*$Tv|>WOLePFU4%?Ho_rb5_kyCubRc~3{^uU(KF+42j%fb_S(2A~EVKYB%g&nj z%jR$C)57$W8Yqip%vir5SUy*567gUmht$s1om7Eks*yQO13KiowkHl% z^I3c__WA4-OAcFpghgw0W9iRLN0v{6u_uMO6A8bY{cLM&aUP!iMD9O1XK|1*{>h~4 zDRBxsuGlVOf z?k{30?HwJnyAzkqJwJRwIUb)E%o*$DBYOa2kW}&xy20j_)>%77_S`YCuRKYY0}EPy z7eR~!?Jdg@Gy46-8K=vj4W*J#lC)F|8xy|5-V`@1DbILFp_Iqv@~Ed|Wx1UyC=OBO zM4S`#$ERp1ilR}y?0);Zs>cQLXDpgvt*)O%g=p6iK&N8Oh>fX7QBpja2km(7loO?S z;-mo)i{x2vbf)v2c^%+?6tbEb&He0FCn4uMaki1QpKmH{%fR%pAZjvcYSjJZEIhoC zlSeBdGtfwuRIh`nq*o;boBD8_f8!mp$f)_*hr3#$WTYIfxlae2y?0nX9Cez#X7fN! zH!u2?kHP@8C2*}+_j{f*Lr^EPSp}jV7YpTkn+fU#oA5_(6(1%Z(wQ~pLUwO&$%YmY z?P*D?+mVHO3x4 z<@P#%N82u7Z+0z`;Pfv2*6FtP9M=Gprs{hbNrum^sfQtBcPNkIZP2D9g(ZSvu~|zZ zFg4XrNF9VV@5~1?Y#i2xN6N>vzb{pCz)r4t>dxyR24fpdDBC1Jc!9adv@j>o=|(7c zcJm{q$|PRNP)oM`DlFG(0lzrZ$dstciSqz;OmONE1G7!Awz*$tOZsVddYY2l8&cBa zisV?`YqRnp@TZ@9;|A0E*p1j;vzDwpL7Gk%whzCzRAzXm*qx#wY>fz6)~dqp$Q3*z^#Cyfaoy(Zw>G~r{o2uN4t!-qp8Mb&j=)X zTc}lsZ*VksVGk@i$%2^e&ctqL6@M)tdk=>q6GpAOw%(C>v`le4*(~3t`(pc((mGE6 zK(xYpaKDzVYS_N;B?|_M>ZI0^z^dKdRSbD>IC45;rF)PmGGGGJWOr}17?Cw9NtE2; z($sc;vnma&7L!+Iy8n(cP^#Wq=MzOD)`3n(^pPyzbf}hhjbl*%Xd;`Uz1Z-+1!W%g zIObe!T3n&|l_tw*PZpEujrctkZSj}gE>eH5U*S}z$uBaO?Zs>>1V$Cos`q-U&!bM& zx1N*!GCKa2^<@h(Z&7Df>+LLD2VcRLB@aT|o(y|Vt?W=X4D49*J)hWg^t`bte;UvIZS9X`c@|e#}vS`#*hzLvY;i3>$EQ^cybNn3CSRNJ>be+Kz^A{gWpGj~1GELbWk(5Ym0wVB6>1xX=jx~^vx94-z zlSMkL>Brn?4(+q0{#Yw({V*+7?6qc$>E;$@Q7g3coxRt$_etAKLD5fE{s&8RTl^RO zC;PnyW|-u=$2_NT8#wQhbC?K3f{VuBx=(mVZt7_L{IreU82*PYfF(t50s7t2BCV{` z+fwyoWBojeST;-~Z;H&_dpV(|jOGjng^!GaS>Kgiu_167hxq;k?KK@9^?REO4j*FF zW8A`S?6|A?BUHJJJbc(VO0-rnrsyJaN_}d{?^VEEWYeDyQyMN)YVS?}Wskly%U%s^ znvaZCcKmyCBOSFpV_1H#SG;IC*qwt5uC`vRS)%ZwDfF8zK59gH)537 zH7##g?Aj{>4Ba+kGY^$ag{Q7*QkEnFbMCp)d_(0+tYbUm01y33>pdQ8*ZOy)oX-Jvl;$)Uw>zRm*HzTO%|G{E*fxe{-oTKN&0JJag^RkSO^Ww z2W*&$8F98h6W!5K#oZ=s>zkQ2_+$CbuCA)E^mq!#V3K>KO;K>Hwd5n7G&ysGz4x%+ zg!q@;G3-6E*OMD-`=Bb&R&Tfa-kM7j@H60L{!fckgF}PWI+92$QChG-NXkk?ibJZ0 z5lj)X4I426O$;sr`^0J=JG@e!z!B-h)a*zGbmGI#S)ibscZIGY@~(=(uCvugI)-3x zI7}b@4>z3DE6>~~-w)o}JGOX==gL~qU^rJ_PT9I|V;Wjs+NFY7WL*r1wr@xUuA;qo z8;A7q7BS1;)hui|VG<2VWc0W%aOzBbHKhpuzWoJLj&Oy&e1Zcyjf-pb`PENlX_3$J zmXr9qrJSI!!!~;JpA>SL-*MN1LQO%dS#qRS!WQNa73bLTtaYA~4|n2`bBeYB3)Py% zb3KrY8a;q!iXmAPmF{oQ!*8>jpB{)|zsI=@7EO6YjtQ&#JT3(GG2lCy{ZIM0ejKr~ z_odT}5N{Zeez9;>H(w$qb3AXSY&zaVRp<&&j`Lrqcb68T6F|=?ItMule~Iy5fz$B{ z#_A)kLl5e389o|kjP3z%JbgsS#-GCfEdhrhC@^R2KdkMW3MnJ#&lOzUngr+$N{oO= zO@ZpZ9bB2N`wVYtdgO|4AA}b5!@%+LJHFg+EAf4B0K=1I!V5m-XYJh`2FuWiN(TMR zMts9p+8|_q>Ht1V%q7pf>F$T$qpL)K=QD=u)+XGEMZH^RqaR$>%d0xi+3b z4G0;BM!SDbe+)!I2^Y?4tsHsHO-}@Vv5J>vp^1f4LU;N#amx>_aQHe%%$31rSICKc7R}^(^1JV>t zjbP1ZGi%KqDXdM2CidNQ8h9IY$CUf}a30$GaTADTVtLsJwX@&9{9;Nr5b#HGSPSag z^gYn6t>SuURu*+wdaXX#@b261#j~U&ciW>gqHIJX`I2C^&o`f2nJ*WA=0tQ<*er;b{n%#!q@j)=_tLD+xzS%{F%}224oqf z?ZafM)+Uc|*D>Z%)f-Q`L_>2GRWgcwmrWvrHTCVc#vXdnt6%Ueg+W0Yi!3*TmX6a7 zO;0m#`Z!y^JQee%Z4M~-8vv?YL1)V-K&>~l$U@B<3*r5g)yT1mMoLm>>Meb#Z$e5M zOdPdJ@#MGe@=>E!ZAxKIXM;&0Kb>XlJ{JIvXj>Uj4XE?WlsYn9(xz zvJ0^MT!ARVXWW9ML(QyBCb%|e-O}ppKc*~u!4?_<*-b;QIv(!E@+0$btT^hswz7R` zCpoqVH4n3TEA-@R#-ho>v1VYdU#DKhLAl*Nr+oiYd9GzF01Dc`(yJ|b=D&%M|EeK3&G;6$Zi(XNasb1q}z-rOeynql5$gCE`1>Wq*m|$%YeL5 zMEyQTcirT?o%-UE>!ByZ`}!EGq5&3=9T&9eGKzSQqWnxbBwsyW4F0$U{*|YEH)12G zz)@^YNfo8rfKGm-4;LIrA_n@^auu0%1c=)Z~lBV(L;-pI5ZR0fV;`u|!Q+W5Hx`k=@2BYOPOu!I!@32BAF)P84dc2{+IH|}BMtQNQM=)H036?(vEIf zSTPD(fgj#+SMDU1Su%i3|NsI(b5!s7ROm;tUj3B#|<~L zR3S?lYAtudw5Faou{aC5JT0e@xXEXz(ZM;#HzDv~aU#FBTWZx=PUJlrj-5Dt5-qu{ zLqgdc+BwVG7yIr!$oz);D$F$#JqA!XWaG%_x_rrzr!ky#6Cjetg*NG9XYW|$zfk=c zyg^CpzraE%Y2?L2f_eVM(?B=MH!aevQo&8TLQ>fjU{_ba-zw-JlhV`$FAM~&#jTv4 zzk8p_z`0>))k=K3Vz>o19N@?U@D%EpvCQT6mYzDk;<^bA$dVk^>P+md;JU2f8TM7` zY*)8eHH=K=%Fcf^N*~Eic(LEQ%kr6Ja2!^T#f8rQV7LBmn=RvN`%;ZXxAh0>Z8Bc3 z$LhRSN^5N<-OU!MJwUTj3io#1tq`<7XYhXZBn$#$(mx5VCpGA%uY35kS0V#OTjG7Z zdRX~5pBLdZ5#NM$+}-z(XfUjT=~dVi1;MN88`^R%6pMAAU^Ak8MC;L5rp$D!gMii<9m>ODTSL*k8rly zW)qi8_I0skl2i3upKY_rNZ#SYx)ZS@nLiqGuoUzi7Z4Ioja!CQfcV|NJYG*vO^f)5 zT>KWP7H3oji?59nv!Vw@b36A9sCrg@?+?rRMAv@QSp?EX^>Olqx0@^k#zi0q$H}Jq2n#sGhIa%B)s2S zVcyf+D|g>2y_G>?Wix^uctY`c-f0kB1F5C>ah?3?vQq68*dqBW;^HRtyxmPQ|Ax%! z^QM|OZG|bK$#$ZE5up2*V6e?y1)AOZ=B>k@eQY+qvv;-FQ{PbF?{GKEb!wN%=|ZUM z4;c6PSEK{ph~@GCbr8Vh5UudOCFsI>WhVFuwu17>z1r$lVa51*fhVsqila@!_hSan zvI1*Za;0Nc+gH1wuR&c9p7VqYbe<{i2zhe~KvjVx2_kd}+mQ1a1HKvwkX_LR&MyPS z{CvbkGNm=Iy*x5g#W~=j)`}@`lH>{MJ7dP2cP+|sWAcsgRiElENn|yl@ORno#n{~T zmdrh1d4T&G+TX`_iK!%g*nHCxTp*IVKZgtiBPBpko$9=wH8G888d%xmJ)P8<(zY%A{7bxN#(U95*X`#mnRDgcq3whli>o1Auk(XEG$?@3N(?Cxy3Rny`nhaDp{14 zh3&6dn^TqELY&WC-rv<9aKUD-yO)*}UdiO+ZlPuvH0H#3y*O4FuinFtRAV{pBZ)d5 zNqQE*)udo@3p%`6Fpp~f^DKBW@*;OJw9=zkT6;Sk*&adsL%5(OMO&3Wi?SvB%P)og z(;8le{Co+akKTnbtbH4%iP<*Rm{rZaiGAj(1E0iH_}rcZxo^D&u0s|0wUtb>{V)3y z8(1NC8}mme;{^4&msCi$;**W{LtwO%cj4g~_D(M+-rvgYI*bwrle<3@hi7Hu;+}r= zvHlVpwm0*7d?af+Tqf2PRVib&*w!Fw2i+_uEqki3YgM_wy%S}9yEi1MM@r^b(luu_ zyAaGGkZ>!B8@5H*{%ZrKvJhJoxgm7xGazgk@Z=z{0$0=4we5i0<*oJMtl(U$ygLwS zZ#>Y4^sxQe;@p-YccG5SGUBV9K}?_9rxl+WXHQ~y3bhE;hmWo9H#jDItsaPsX3Xrq z&2RZSL*nDN8^k5lj0{Mg6a@(}Z+YhIJ+FK{4bS*%F-+L48+=d6f~qVI5vS){io;@9 z+=UgoR1M3Vz|kK;Vp}JnA%%mA-`n_OSPhm6yVO3VH?1MdfgnLMxT1{5vwSBW<^9!d zdO!L&!?rL5&xUYS4a*sfRu^(6o3dA=GQa-N#6rZ&oK9Ny``G`XU16sw zBvB|zaGKVn+&M8P;J3!eb{Sf+ZL4sP;YedKsih{S^H1j|6h_VbPa#vrSdgc_$>rw5 zN!t8i*|O1SUE*h78yg#}p(suhI}D7@yy{u{ewEY7qpyxak`(!9nk!fPkpGY+l{6oI zF-aKfE)4FRM!2rHYZB&dInjoBeacS#_GRlsr`pg%y7LDs$6@V~`b&xTW~9UI4~d1e z;!IdlJ<*L3?ZWGcdogW@vHUoxnS77)=+P2{(eHygrsQ$nE)aZWHp^Zs+k(JCpgx$u1;2ga|*Fe7dgBaCHA zYkYviGjIITOY<|&R!GcuPqZ=D{KM-wxkqfsgfn_YFpj7p_}-b5_=HjF1p`s*N@&k~ zK=RdlHum+Y->O32pQ3@)2!t3qki#gYho(I5i9Ci8;u`;7ZtP;BF?K6(! zX~L0oy0mwv0MXYv2QR$%;(K?+cA(+3Vv1to5`QP3{@tE1#sl#v(q*g7&ri9=*XZo? z--IMwuv$s(btbn77l>|T24gHU3iYyEvPy$ON%7o^Sh+CBgSNd{{FB9*aZTd`$7fn% zGWQcO4?CV?JN}S>({ua`x1r0zzACW1+PJaKztNzp7QK89cEUto(kA*PUuyfWTP!Et zOXvzu+%k=x z=;IyozP89hXw>4E{gK?}FX*u{4?)%|AK$NSC|gg6TF3)Mhqe%ZbrCb(J33&&Pl-`7 zAm#Ll6x(wkpQzG-$K>l@vMRI)OZa|D!v%8w;qcO?^!u!$lWl^0S0rW#zT~hvP@HW+ zI1ByO%a0OPHf(IRXctH_r~Ee4XXu9_k-4IUaWxWh7M^sLfFWTm(G>aj$L9{N7>dFT z1d|ka$pU1*yV>iM>8Y?D5hCbo3=0Z5O3r*3@Uw^jnxJ^wK6J4C=uHmkbTqZcVAUkK zyNhsEbxoo1-e>xwIQW?b37He+-50|^$BOdjge71}=kgCv`U_Y@N274bu@8e&P!+|&hxhYf`6aoj zvTi7Q#!sE7ItXWgxxA2j#d#3~5xXpXl%w7)_YtNGx9x~~z7b4>Dd88|aOLj;7oOzx zsO#Xygm#)V@ZYEzhA3xA%Atu=_xu+R+&Owntc<5BAX@ZIj%Ro_fMK*9>|e1(bTd1* z>wu80rxL16%9|re1j2kVi2g@=P{IG@YC~<$HLZi-p;pHSVMI*~M=>-Cs!VsaZxz(5 zRE{Ld4^?&4`sOl1-isq`$BMN#ZJPkheSUSA^4szUW1LGBgHiof==xxH7=9wtOOEzL z)1*zJZyHiObbOS5xE^+VIr02>p@|6XD|^$74c|oBjc!(|dLP5YQ%Ev_Z~!etSk{5} z5>6imCvR6fVd#J-vknwxd-vKAVI3J z;0r%|) zf6@_b>gP_PIpvFiq;w`h|1w2|H%!?@Uz*$SdjWRBlt8^{7Vm+|UcP@*mca(5TB-zL zp;_Ye-_i1EP+k0CB3Liz2Lc=XQ4|T0JB1=ccwZ2(y%17-M@%m~^#VcX9%3MF?J1BC zS#V1tGTcR|7)V1~fo*WBO6Yr!9&DhNDM?==+AJ^)9>9wLTQi6oYK`Ks!}hX)kKFkd zZNhbOfQOA3?xGo*%V=e7_QFvRltzr)$ufUiERj;U5}Eo2;(!5fnBD&_a&T&6nJoDC z`H~n2gKwpf_O~8?$v5^QZE8voJgnm&Qv@8a#sZq047roR8kdvDK*|KY7YG*^Ni}A& zq{&E(4nTu}C4y5jz5AT);XwZ8h4~G#KK`c&vqt?T;|D!>>{hO<{egJ;vDAe_3SrD)>9E8>vOrTXXB3O6~9}3Df{k4b+M`Q zi~B(j9oy@Erv(F&0tde-{Ae|M(LzRb0U5PtLj-;Q)>|TGxC?UpBnC>PD(NHsCJzgn z3s3Sl(g&)Gh`4dV2R%aU2lWwprxw2lElorZ|@6cQ>kw(YQX5 z2R$v)!TEXD*V@FMI9+`&Ckuz^a^ zk~qH|)1jQHrKBcujm?Ec_eQDq@);?kvGMkeOB`>CR$0JojTTnI#v*B4nIu~zHDD#Cw?)5K=bSU+eek5|Jau5G5;5Gzyj4gtw7pRc=f+-6nIYg7fH|tRcjHM znesh1(!}~7aD&lYPo;@B_h&C%xg`Gr5WX0zY5Ah0|D(E33d5d?VSkT z!t_D;Ewy6lcNzb~r%x5S>_3#K4Z}1IJ1s6c)Gz~63BM@~`@euHRy?bvS3+Tp9+FUO z0W&+}UmU{Uk1B)~3UctTe{?m_pnUlfbj^hX^x!v80WRYPOmupm$O9ZL)Pe^}+WWI8G%qV1a-Daboy{Zp)n zEr;e${9IXyqA>Imet{gHz4A};50z1VXSr-x)khXf&<4AXx#y0uWMC#^Pcba}({+3m zP#5Pb(Z5gs-yD$~=JNj^L=_0Xx*{h`FO??wIrSVY^a8$Ht)q(ttkv^be@KqoqJ}kk%?daEhnA#F2TiIHaFX}!1IuR9# zswj`b38?UjptF~ttrnJkvNIzMwG$(Cdy6w8RMO5yxzFeEZd6XNEx$}RK$G+3K3oHk z4tfn<&+hEK0^FI8@493xVFK}5t9PZ($6sz@pTMEZ(=1x4R7sHJ9Zsu>Ty&NhaQODv-?v%;?D4k&x zvhc9eAU@YbpWdz1Uh40@G&smYa3&?u9|Z}tonmQV1NW&|_mBpR25hJt48~96C8;jR zc0wa>fS;MMiJ_kz;|Z+KfVhH?EOGh2L?@x`%Ih7J`;Jy zg!$IR&TRbvq~|Ra<6Qk^Mh4Z0oHAUv1(qbQ_k(F_*OyUBG9-w>x(+8R;>lfIQk|Jo=EYK*AM0wh4FhEtNEcoSbW&SGxn*QIgSRFgd zRvH>zuY>1D07}68>v`-s1^KJ*-KEk_vtyF|`BVk#7SAhXGeXgu!21#R#mmj8WdhJxo+_~R36&+`52 z0p$5i_O{Pis0-k~DPU{oJ6jI?`cf-j>T_Y{|C+l2_~b1HB6rz{ZCVb>XI&d?Ipy;a zjv8ME1ETFkH8eM~?@5i&1p|M5>lNJ5owrwRoK+^5!N|Q8j@`Z~`7zys-KtD=f!Suy zA_rc2-G^AP=9OQ8YefB)vR5*Im_LlTfILKXPx#Kb`gGV{vHFe6*&>j-$7jA2#m4Qg zzS-WL97PVcW0--FNU-M#%ak6W-Bj>rQQ(8j<*uZs=EUA%`*%*4VxVl7S8&6bV*GoH zad!>(uOYHBFfecI|Na-C|CDC8m08-@8Cs3&r;ID zd8qScs=qwX_f2_RE~yG3+HlpWd9q?SwZEmWqjFjm0||48oW00i_gp^iH4F{?F24zS z1x7c=*oRr=z57cyxPqeZFZUcawjTXo6VJS(pSAW+finH(8EbdW_oB%9P>JNf%6|x6 aDU;F3TMbS7egFQFLRLyqvRd5u`~L;;VJQj# literal 21926 zcmdSB1y>wR6DUj;SX_d;1wwFl3+@u!Sv)wwHLz$1n&1w>La;>_XMx}r9D*$p++}h2 zc;0iqbMIfcJ=Ha5s;Yasds?QdYbHiZLlFm)0uu=d2}fB;P8$gc8TJy9(2-v#%kU@a zm(!7!ny&oI83{=g3CRvgR1`_{1t%fdHN9MmBH8^%_%GOr+Py%M=nFmR1(xhyr2ohM zA0z2M=6@mS<*-Y7!A*8AGEGf(O)vO42ni`k^aVtd>|Wr1{G^xA^l~IMz2K#!rKY5% z7ySG~ORLEasmV^XDd`35{^P$${4X{&Ej2B@z;n~{b4yUrKxXDrlIT*B-BQyFykzh{ z_J1NT;d$x#dAYs)xykN%Dd~Bs>G_59U%D6kjNg^z{Zd^FcO@fFBqXxz|KclU@9-2P zBswHzIcZ(LoTFT0WF}~0lDFON{S9Icde*66y z*9hA4&3^c2QK_e0G@Uein$EywH^H*ZA4f^^$tPv}9aU39^-Ff!gsjQy_go?O7Qr-9 z=-E7u94;(JOby%RnJXNrVKCB(CqP$l)9Ap^-+OPzeG5!l^;!?fv4+ob&a;n40Z+S| z7CTkdPiv|SY&^@XfQn4)wdaYacSh%VI!+F%(jbn)JR`2tSyCuI84(@i=NiB4efe$Y z-pG3K)bEU4eI-7__s&dmRSthuV@RCle~2-iX#P}dK&RF|(ERP)nwB(pXc~V(fCLDy zMgrfv-;X~ZlwSOL%%LtlWRBvB_7A+ex>T0hi5^-*`<=h7tm!OZ?saP^dS9R7fos#F zNbUVe4?Q8^ZBdlLumnwx@rX(1H_0>tg3J{hHa~RiRsc$;*+X;Sb$u6t&;wn@f}-=V z?2aJ1ALHRZKd)vT@|@l|3p()Q*ezp8!lA_x!FkToCG+}rOlP#hz)LQmI{~eIX|G#_ zWy&FptI-H;y3;qHc97Qy02>#B+P~==B6;38cZpI# z+#!E79VArGmS<8VI3SRq{3KL;-a-zmS>wMRoDoVth3o+TYu&!-N`vPaO{Fto-AC=* z^$sn!JPg{O_8HMtC`Rc>Tqt6BFx|JnSm8!Hj0eeYi!a063KND$C^+PEv7=85#1?O=D8;MBmW#T$tQgW;I zHEKr2*C+%aBMJY%ygs~ikB$+v0!JT`pAH~y{EridX3?2He>ubgl2LKF&mh$3xZeR-j??Je8$y@J zu))%=m8Lr)@Tn|N(Xly!@{0iMC|KnNJ@@Vh=6h+Bgs~nR6d-EB*wu=x`&Z2OAb0v* zoem_h&erSv7-V#8M0CcUBEBD1I6e0QwPW5%N0f$6P$WR=VQ+dsQ}RB~oxWpC6d4=o2RYL z7;~9kI2KH>r!D@o`~_zwOx^Qp@!GArrjty{&$FY|nQiqG`Os|g5lT&3fpBFwvUkYnW7{QWToIYGrb{~ik)uHrvjTu&DJY1@;}Q(up$p4K1DRB z%VV+#g<|F@d|%;5GbJ;nu70)TAm@~&l#R9gvsq}JGu_hNMR>qT2<-b#AP$qky^PIS zHT{<(o!Ieu7>sitzj>2py9b`-jz@yx^C2Fl26C5&p!hH;YwCIKegw?V-lyZ(|MVg7 zyWkC1_G3;{dFc2f4TGzIaEMTRlJ+$!)BDh|!`Z+BYGQG=W*5wvQoiaJs}_t;6wlnU zWZvAyCZvP3dR)c>dHNv z4`76;IK6${w0YDhs}1c+39JwYDhI6D}okIM!T z$4}`<@{^)-XT{}k7ki35CB{AZIzf(!?)i%Jb7BjPc)i_pe>gI>logNN?C zZ3sy2bSRR+aMBUB#o@|?qwd?WnV=t<1!;pFmS#TiRAmgk^kH5G{$-^~Ogu0i6Qo9Uz(1Qv*mc_BTf$-HR z{)Zj;CdN3=)mlHEZFdvh`I9pRSwaVOE-r9@fO)_U;HITwmB0u&YqPY;xI@6;Gjc$v zi`5$x+;@_xF#Ybkv+L9J6l$O%J_j9&;3sUb*{eEQiq1|6FMQl`N+taF^pKdBT<<^Q zFL0@n<;0<2!8w+Hom0*6p!obE<`vB97RJSW6G=vik<$@(NhJ?jnzzRb2cq!^W3c=V zg}qIngY;HpYk@7@F~3);K@C@B;7J>ss6N%sfjGTykrKi;m5XUm1l_U0E^CPdzyw_E z1V+8xQ8$C!9mqcELdRzsB5Z0JGAemqHo@1zo)|_y%|PZ4M?Yk^v2mHqbi-jlR?>_k zx0@Rn97Is9QqSeiW+qMh9y9{Ay1}NUTh=32(qgPvS)9HYvx!GRyD5A&i;uF*;({roO0!-w!HKpX^@Ud~}YG31Ll-hxhD9Y#hUI>V{ zSIUK>VY30aY*nz3=grswd-I12i;#aLwDgetu1~Tk;S$Q6a-YGTcKL~3fMo6||W z_Ki)Hrl1O`PFgl@?CoU90$bc~_%~$-^>GF}r-8UVy;5B^#I&3cd!z*FH)My>xJz{0 z%B($F-FDAwN6^}^x_u7JVShw0aRkfX7<=^B#dno$e zvr_c2c!a`(YbZ`1$wujtv0G5b)BW{-#lb9}`tb;D-F2g)8I!S%!ngXcf+T^TL)xt2 z??X1DG3Xc61t;ma<>T!ObD0a&*Gllo=od6P-(!GTzMG+S8dM_-SwLRyx^RI`v4SlKJ-dw+AX{r1WePCH!csv9hPqGkM*7m)ZX$64(5_ zuQCub_9LZ;tWj>agwU)+Jkh4xSCLSB=b~{Z3F4XpC~5PDvt`NtHC=i`0b#+B-lX{m zR?WnJrA_G0SblM|`06~mDm>`jY^kfq=-}SiYr=(F^l-LtMtT5{aF7TAU8U_6}icEMO4)CDlMHXu>QLnp6L=9HTvI-RnvpBTDZvU_zvo4%$Y z$?+sI5FpBGxYq#8KHR9L`+MvT7M1@ihqt*f^lk!5D&=jK!hKnI3UjTiQLJ5wUlXjB zG|5goo&-B>Q6}6OU_a>1Mu0gQgks2<$L9VW8J=fDsLm&>Ts>PC3B)&K{%2bpxXK69 zK2rndp1|Rxna=I_VW@&e(RM6F$HZP7^I+MvZs{a#Py{?voqdqW{Z0*9gbOxfhKN6? zHxHy@cAny6gMEJh!7E&!u37n3zI4MgUFHgnRKTzgeV(X-pZAX3hF^zw6H*!>fdyFc zg=@Q%6|tkc-&|pu-ilOWqG%ksP0`Du(<}^!i37^UJ#H;MXE4X_o9xfSW*+FPm+yGPVagZDBLA+T`rx96 zF^1DViuOIDf|`9J$3ERYBdE|q)mtRJOYlz?{^@Usn*xRjD97bFzj~Vh6>>GE&aSjCNc!y zYDpj3_GQ5QhpL7ppnU$9`3pxL5eGL9Z#dH#u6?F2DOuQcbF#2cG-t88S@16c(x5kJ zJiJ1~#PeEub#e(<*xi&^ROQua>gf%4j_Mm$*va2f=X^i@J_!IKw5svN^-r{APMg?v z$?S3bOMd;$DM`5{c}nAeV0u3aK@!hL+g17eR6}&IzRA|?HiD&yF`WHgc^L}XW(ANo zeOXn5>qVu=m+Hjvlg@*g*ty*CcB_H_7TV~mDj?}E#qtW-3bAef1`XaouF%LH0y*X< zm*gYi@wi|nHUbh6U8#kZjX@Hp{ z5I0^Q`}dg1aw@(625Ce0)<@Y+O{-;Bm@L2llG6giByau6a`-|`<JJ|NvAwI`9zrEsO&#H z(LkH;(-M$?UGW7yqO9K1NSl?ztH z>!z$M7-K4XM<+-boK=3j4TXwT!l<0iHM)zU;VS_5AD?2!X6_!o z682pDX3eDzc|1HyVC7zevzKU6iJ+X2h3a5X)!!O-Q-`0xq*1qxk}cpxQSi!G()ooW zu2bXQ+$)J{7<;(qTCDpretTKhJ(g)w*8p(_;kAdwi1@L8OI#m*MjN=aL~RNu zkU6Et8E@I~yw(frDWZVPK#)T4*trL?4)JEFF1h{Yk#dyar}3)i*Mk7JdM;o0QT%o% zB#qb4Iw)|42Z;vQQVwpd=b*e-Kk=o40p5#t{cI4`wTJroX1XjfKL{ZKy{QDac}Vwk zzs4D-3>A7Og}`rSds%H>tZz?S1M9RQZLUC?;9fS#8kos40o;n_Bl;+)h|+z-;7`d- zwi>+b1RJoH8y|GtyVKEUQTKI%9dcjVF6efvh~4dP@TVQ~H!VE$orUynwiL3mxVVd zF@o5`ug#?u=jjBINVjU#L3?;;^{@ky(ke>gXHl#_Q;tBcy)Q?5aTX2{1|HR@>qI2Jq-~CzS0y+PB=YCg9|c5Gx+0#$(swKA|7o z(jr8m-n5y(rzaO7a5QTgQ7G!mdyP}~8KIBh2+wKVP;jyj#u)c(L%AI!Nq=S3&JEHR zC-3EzF##;vBpe!D>*z}*G#}IX^vcjwCglF+3{Cywcr%kZpAS37@4msITJrvB>}r;D z6K!YZm10O6tbBW9Bfhizd*tHOyO{Xcj|b)q<}pg*EAc6dn@De}O?re7;(%2vB#r@? zq_A96!T$fKalTusK%GFgwmZ_5-+b;4hLt()3*t3c?MjRp1xu6I# zLDKm1N_y}~{9&E%^NgPV%{^`Y1VGA#<`Kr4M+*)=m@PV9nKzx!Y0-~TV#g8XaI*L& zL)D}y`vX!so>iVzOQNu!^q03Nfqm=nyjtln8xVt3>c)@i#;XEqGAT3;wyQ*9w?heut+$b9||HUPbjAf0f$$-?sRZCN4&Eyv=BVu zvd?!lf?COubk)Lsv->yF#Fw%}$F)wye0-*?n^IU$gIO=ZJ<@+PCU&^-#@BPZ*H3L7 zDSL5wxt~#Ds%a(<#THx3ex*14jNZyDs&6@L{OtTgEQWKcP4)*0?%$tsfo+@UK@9ZT zL$sCJjFhxR`SpC2-Pg*IyuitHB8=%xuz&a zhOfNMh93++Zc!y{*0K3{?G_|W5kT!Eh* z=i2npv3wm!CMCXin=C9vVmb5gKP>*iN^u(h)u6pf5@h_|rYpVt*6csr!-CHnsRkt^oDoywSJTc4FdZqtE*`?a=qHIwga{F z12w9kUI?uxvu=oM_V#4Ri3R%O>G$g;%x`@s45t-WAiqPmi$*f51N7jlBKXhAJmOWx zJ0G(QSlX4d<{BD;$g5P=?2ye)#>zw}6K%F3Y;|@S6nvZA-sJV&)=+Z`Ga$qljVL7Qx`$KB zOZT!Wr8~}2l_x`vX7Zyi+V?29#<@^icj41t?VSE-=ao`l6{V}nQ3lA~a#{89C(FFvOfRKQy2I=&I~osXn> zMC8jX$PsH_NvOEd>1VoRJzk|dimCQ)E<_HLPb-AQ`Pm8FXaCA8{7O1=z7*r167);n z+trFHY{XEWua%?gqp7DArHc8<-!IvNnJw2NBbcrHzG&N1QlqO!TaYcj1lis!@64+4t@XpPKmdfH{RlInnZehxOCvR3`p++9$v7 ztcaSs=z=PJt-ODA9iTX-Giwx+&wj<)^=eGg{1`QkUaq}?TJp2CMP;jGE7GF;NOk+7 z$tSQ4A3P8Ia>u)GSlWjC+Z*mH)zNvkmY+8}OdBdhVtD17DSg3Rz1grY7M!mYmyJSi zm>Xz9cDTdET)1#qBK@m-a+b%rJEl{IDRwr!IrV($->05L-;f+`p4P5&<&@j-WOqj| zhL-LC>h?be^_}k%33bA@-_&DeOhR=`Bn^V~p`JDbET_da7YqhN2%njbKV2c$-9{C@ zLjE;@6kwCrp1eI4lW>0^Igq`iTj*Qt$JUabgWJI>4Pek$+hVJO3V|R&M~trVefRx0 zZuF`Dn!&|Efnw6|=#wY!@Glj_w9M{K3aYAiVwB5Y=f8f8z92solT{OAjqQHxA8hxo z0{LbgI*meur&5OP2js(=5;XkLQQK*?mU2V)Ji0OAx6Qgej}iVg5r4so%a-S;#+?n^ zfWcEsNXc%{ZEcd=wQ~1eaxXBssuQ#R(Yus@wbWTePJSKP$*yy=_A-`#N|#5bCg&~| zX5Qx<%YI67?K;?phUt7c^O1)1EhmM2fvg2!-=?|B#EWi=*OUf zu7gFp@=u(R$S8TwTrMo>4A+W#uMCO&Izt;A&ZGg;^m9oz@YIbO#v2<*TC`xzG;s1HQpcs)9QAo018sP)qP0r3@49wE6N5;uu;%&6(Efx= z(TD9^!v{+fYO{$y+VrioZ$))%t;7%%H1F!eSRl1Q(nFJ=sP?)6jYw)7-x@zXjp=aM zIr5z5!T;|Ikf!U!Z%fI*y%Js{*0Wj-_g&PE9CAn zz>nop?TOEr3hZFPkwpIPX_!UBE?n=wNBLK#rkOs4Q#wNrqq=9#F7SX21@wciobjTw z+W_Q&>d8B_^o{Lfehd^fZB5~Cv0nvjb$_a3XxBs$C8b~JTTdswA!nLoMsCHPNG^J0 z%z)@$;dx3p2b4+q#VZ?W!Q$ z%m$Sv=fRJ@hY~76e9U!xD44S9RlnQM#4yb<8CRG%fvCri#F6t1aX1Kb3B0|dF{Zn8 z&t_+PW%X9Jl)482jsgcaJ(sc53bOY@^WdTYx$|_sZ!58e>z_*Gfb}E!U*D_?E%yfLuc$xoW8kmC`UC4cF8RXM#oj~FrDbv3y77uXh7&P!j4v+1`rCQ^w_+4GT z96;7CS+!LeiYg;({qOEa&>?kFE_>EFY`@=-O!C4Nue{gPKKBNd(R0R> zv>bRe+j9Is+j=*;7#8JC)PAR`-{Z}`C~|Fz+b>gWeY);hQ(hT|Ur{uc%Z4qS`+lq8 z?m(#pp__lLj1B{zQvkHIzjoq|I=6kW@?QN*jNmqm>8Vsl|JiF)rCr=99@usz*`O3I z)*V{N`?=nToYGFn zpHzv6aVr~qva7To#x}V0`0J~~de(8TXYp09GAHgXcMx@rZf3?+v(z57cm$KuvP`{) zE==Nr*r3Gxw*;}8w<^>n&?7$m=ejG5ZZm>#7v>*suCBZ7!G}lW^~bCZ@K+OWiHd9wDhQ_<()-N4t>SM!-q92l(Jp!{{%+MT@H_MZ!dn-^+vS4O$?TG zCw1c{4og3f52)_!_O%&eUbd`#Ti4a}ue}?|0&KE0-|FBIS-NQ<`}XoF=XC(}v|}3@ z<(K(k8u*TUOdo#{bm>-~I{$}$_)=2V{(P92v<1@$U9c?%fO#RSd!{_AD{DXjt%n4^ zDPX9GW7hibUrxy7H(}qwJ2Cs|QW_^NAFjO}`EW7#Y!J^xGo^rUQx`);PgRH>DtO}g z5KHKWk|bF1U3l=F&&q1MsUE+eZ-*PD^CRoL6Q@V;-h9Ka{d&lWYlEpd9<=72OM(5E zIHFC-`PLzICB&S+5-%ctS(W}5^pWWsHiD{G8MOYPVBI>rE05SDiT5Jk^kQ4rf^~u$ zr?COi>dVItoUn+A!~{2IkLKbfjC$=$zHk4+Pppveaf$N{S6yaz8m-XM^E^h!wN1eJ z{4j)fn7n_wW;%L<>fTb8a7z){t$)Y{v&%e{VA3wbVnV|tJK#g}h)zR#bzO+qJLzrQ zOj#%9BjL0FjqSfE$7yLS2rQvbWWoHqZAmL@Qz7Ol4&XdFwM%141J>v(qKS_XPai^J z>v!Ti&1!dhUAG{=m~AKNS_A?=Ic~o3VlMY_<7637ShN@EsHNJETX>i)X>q5NmeFZ% z>{$6dsB+l>7v`w=dKOdET_u_cKze!cLxZVUS!q30F@@wukV@> z*G#MEBsEuB%<`Ba>IYh?Iw|ohn)b0zUy0KN5Y9vUEH0-I1${hj^RDeUXqHMdui#x^ zkCC-S znwl&uf^mlI#dh|~SK!G0WTHhgl@!p8iqSo6PH%Y~s1S0m`T9=(rH5|&HwC$@W_`Y+Zm>hm zpYcHIs8$>GnKH`a;>g{=_Iy%3@6v*PcZ}|X#99*{qLeh`GQx`=5X);mF5>j%saq$k z?u5IJ7d;12X~u$x`c1L4-1K$h>n!uW+hLJ_fQ}i9=HnjFAy)fRVs#=`^wBJ$NWqhu z*dH5*CfRQ}v*ceh&Az&~(r(QOTX6&hr(1c!_m2w8WYpVo#;VCNt^&-YQ{xA$*VVy8 zsm%`RA1!&-p{frTKmO0oPRnr_$%3Sk@yrZU-hg}_sX590vI&bV)Y<#TeCA|?M%$uA z$E=rly@KT)UUVI%F*G5&DXQ2rzKoCZ@IWN2Q+sw!L_4mNC12t0i0#alM(vpp%bAiX zxI;{qvtm%WFJ<8ZvFn|up3M&6;{y>2Ui5#Sq;N4x$O?0JPA)djon@|ekA3Rgf2Q_e zPU^&?HZe}=+CzCN4|y{&EAo&RetN1oM-FKKWT)~DixrK@FPVC(96pBjdQc9UHXYZ@ zf5-tBv_MH11sC-&e$@~jCX?``&t4E%pvr|<+zJ@c=-D#%1W2%zKlppDh~EV~UhLSn zE7E_*bUvSFI*GH1@~m5=M)3Ph`HsiEam(m?1yq%Jcq5`LWr&Gwr&V zElDYK`S;tn$j_6J{AX5;b={|BRv`_kcAMPa8Z3}*i_(raVI(<=cApkJ)@6%`AF28H z)>82uYN4>;^v-m8VGf127G9(G>Phsk(<>LphUOv@G~HhX7D)P7S&A~6BK9^Oeq8aSAv8}< zq&NubZxZLZhMI!ygsNm=edcE7rM-^mlK%O1NtM|yM-ka?B8R_N?kNQ1=}S-{f@{Mk zRY^Q?jT8S=+0NKwyX0hi5rW0dYoE-rv?k z4^E5)3Zsl|$5P8gXUIbT1Bdx%c&x-3ODrS!VGz0 ziT^ki5YwyA8Bjd@8oaG9TX={MobCDbV*2d`nH*de?KYasH&}%|_!-t1 zc+Jx4YefQe(8LIEs=1o4Pan$UbdC2p(8-y9WL48il^oDUZ2s3uXkp&Mazbfducp1q z0-Nolr>61&`#IH0)N4Od!vUB=yct=mb57-Ke9KS$t=EZeFNsI8?wZG{BuS>1$LQ-27cV?9Pa`>@+BR^gM z?r1}v+)CP0JwhRVRG3nMXzb;)6Xm^BqB_%b)ojS?B<3aRoB2WD6=9|(uk-~4yE5aV zj%DWJp1FTvV{zryCeW^G`JwjVc`R-s4m~&KM7@xxfdEyJm|5R{@o_#nj|aD}ojN-t ze8T&rQYX1toT*w%JnFlDJ0zBL9}6@H*~N~Ty}fr*k|U7|DdCT3*5q7RbCt-0vKJgV zeQP9=0okD~CsFKN8=mdN@1CqmRWXc%%r+f*+7Stze5EcknzNR4*gi{QHN^(u<>3K| z0u8hcze_c+&|#|dGo*3@?=VefJ-$MY>Ha0oSLPyooRHE`eAFr_iB+tfR;(viv4XE| z^YhJvpVCkAc-G2a8?~#0stAiEuKDR=87_@D*)DhS0c_RBMi_X`AJ$eNcwP+F?T2fq!D8+BcV#&G zf8x-ccVj-i_X?c7F!fOiZ=Co-Z}^ovQM*l7gX5&(U8UYT6ZE_f+k65NdLdUth{Hy_ z_^OPzFKZUF^@QUv?@F#?d4A82Y)vtzVXqN%aY0^j&&)0O%v0nJSBs!c z?JeorF+TfRy;{OIOCv6_2-m?p)r)Mggy< z>{t4}1SoKgm~JoRcW$ZmvY(&Eh2*em7mE7tFayR4O#K62jQ-J^YfJI2zt0H_>wLqo zzj6*5itmVHEDw{5)urQ`_fQs2}#~?8A>o^R~HJu}X;H)~UXyQR4b#*fsm{$7U z(I0pPpwicGb_!yC8_H(AK3jq7A4xJf;;{%%Uib3Wjv8ivs%DDBY$vO%;og2Yaxr@J zrew@&3xeyM^sTjI@PwZq9sR;LY?wbA%&PfqZOnm=7puZh&XfOdrJ}xG-(PY2#}~{s zbfH)Tz^O}M+&K@ezUy!1o*;Mcr@&w5HB_4mMR!hm%I7}9ng1M4l_gL)7EGGEHdK$E z?keu=#A?$2MiQS8_|?vroS+9v1Z(TCerR@?)q^C_cHA77#l1P+S2x*b_*|4bFg2f7 zuI}1L_(xgA5CF*`XEUjl&7r`mt4{7yTu*-M&Vlfm602Rk{SeGEw>l3-4ZIZ^&C@l- zDCUZcZ1_=5RKXI+>?uK&77ls%U8%cvn@=vF|BEX38M2~`Z#X(z%Mq+=d1}b(O)gFw zP8#BtQslZ18L4J*|JSmS{C!CCSIX8L|CzBUvlviB`roS>$bCm~`rk3J`9HUN zRn@wXnTwj8e3kE`&W&D99(dtP8dyCW*mX04xqZPdS5aDcXnwl#zl7Vuswcmp9^X$C zP-^DQt~DT<`WL^YU}{*&_mgV+G8=b1r>awL_Imq`$^6QZKpA2D<0nG)XJ27w0wfvE zr1$Lu@i98atA%3fk^J)J&lfg`CWf=cMab>NX}K%v zKvnU|W4&6P1DkDk5tQf>G@Rt9l()IBwjvNLg(-f*W&N8kPGkGF?e1bFmXu5K#p-<8 zH;NXRAGtcVO5?%`?O$T2hV(!_=(06Y-4+^(a4_qAotOJ{b4%1PK?l59`jzj1i4|OO zFHv2vgR0rwRr+B_GTIY+IyZ6Q4#7O=NbgA-D?q%-P#W_o=>OWS-3iH}ADW1@1JjIXh4*ek3Jqyv zkD@C)a(_PnN-vE2n?ubdx16aqYKo#fh~_NpnQ}j&pU(?po7>a()KVe4NxJ|B#3{Ev zg)5eslajz0HqU6S>n(I%L{Qu?+1>_|w)$UvRFHD;3n# z%@<*H#)P#@KpPzmr=g858JR1!3MM-KX{a`1JH$@cR8z0t7SpQ-r3*15*43_{{6c!v z{^%XGs*a1)=xY^RD_tma)`yuxkAl~ZNHP>$$QT`2*9eFi*TR{Ot5;W#ZB?C33?TfK&pEkq1OqX3syo8Hu!?vf2}t#lpmsD||idtibC%`0}5ZNkm`#43q@ zVN%_^C)g|E2v5qjO6uCrt>{%0q5-7Wd6;WG_}?_O+*KB%-T5*)q38W!EDhckUA_}` z#(8p~bX3AK)7!kny@Xz^*sXZ*`aOV5rR$;WgR2@AGU=D%r>k4I-F$+vs3RHEIUKc* z_~Hs69d#U?Ad?WWBSSy@jIPWyFR4H=bZ2WeFtad4pqO*sPJ+%yN!3rp)*M3)Gz~vu z&Y;O=RQi9+O}-pDx|MMhjsz5h*s}w6LKa1jJfWNqhl99Kuj6VefTIM6IOD}noG!%V zIe>+H)NC^VurL)bwzQ;o_Ws$G%zo%CKr_ca{g~TLC-%2%2pZTVkN{bbsqvPkNht0L zdLp}B808|=+|`A@(BDb#XZ5*B&ugK~PsJ*#i`v?CdLd9T@1lMBz-9&D_?A2Ezi~FZyWnB%wn(EN5fe9izcm-T4}4b!4sOd(~e7 zZE^&jq9+hZTG z&I4unHo}Ab3n{6Ny9cf{38(7L<$@a()Ky-TwL$u3Ye^Gl(DJwoY{tvYD5I?@qk?M) zaC#E~$>365tW$N(ik>HhWu`RoBwDun&{5X?@EM~k zki{6@w}#m6_r{6&@sey(^UX?p*I_X9p3NIEKoJJzH586#N+Bl*Pj^}VV0Sw=0s35% zBNO7q^BJ|{rlRXUSyuXP>LpPkT9;cj*Av-7Q~lZy=h>_u^-KV(eVySiSy&($Fk71pCrxUbf@tB<9wIW{EQfVnWAgi^0j)}ZZo^{7i25lPU#pXccx)Y(suJQf~Kzk8#Y zb;3}iKssON-%^R_0F(27(hfn8A$;GW=pfJi#0%Km?`J{A56;uYl?b-bb8wd#u2)?6`6Xi#NuUq83YSs#o2cMK( zXwskfV+zqv7}Be%P{2%b3|VX#dFt}?5Q?;6S8?)AWDgEP68-+cwF z37kCqd46WEdzQui&lzb?;y^G76UB$x@2K{RPsyc$;iu^!?~I2WoeIO?G8GZGS&tdZ zr?Pp>R?`pz@BlML;4#1zzPqLv_`i=`c4!r?kAVsFr%kaRpJplYvy@$pT~E-1)eQ+k z2I^#~UH_r-FRfXnPe{Z{9efwsVkwS3$&0O7_rH&8!Iuh!Eph$0y%xJQeADXSS~mnz z6nPaJ;576KGhYbLVYZSukhUN%-li;em#f~b+jp*o+jTM=~fj=f=e z%1YL4AvR{-peDJdcW0~@nOo=x!ARr>UzfVueMI0{T!i`I0NQFKU8%w(zFv6vc~8?r zo=O_yP}b8|Kivq61L2`Me{AowO+jlLi43P{zzxk zV)Sggb@LNuG~ke(XGT>H02Dz)nID5-8{Y)^xt4cy-$9#}lW(1V5)7<=TEvHhajvdW z0q&hGB4JFC@NoH*2m}cKFn)2v0TI*!#KijNS|gW&+np2JxKkT|LxHgsl#KQUaSOes`IWLnH3j z_WSs&zQrS&>_W{0p~5u*V>8~kt<1g#v|_&WuSP6!Fjtgh9xM*O{m9b&CnM47Z442L5sZrB z7H*-a_lJR)tpp*S5AC}=sIF!Xf&#*&?DN_I=-uO{X`HpUl%|TU?*wZuwGJD@s@D%^ zQBdrLscNHWM2x8I?iec7_W(TyW$8o^V*0be!iTYQ3sp{fi6kvMAv!VHnyd?mbz~${Gx*oo48OUb_h4=>^4qLKAk{>8thLug@CK$B zOi!ARhE4lvo|7V3H29%c_gXd~JA8&{NiSvMTOThT!`7fVOii8|4K(;0gm`-{bSh7k ziUiJt>`d7K?cXUT^yV-Nw_9}l3-9WR=zNt(|I)16J$Y&-+TEX6Uz9){_+9e zZvqdIuwU+LuQLn8VzX-NE}LD01#+0)06~w3v#>&BZWOyso@_yW;o#CEufYB-rVsrN zpP(iNOD+jt{cIh7_KA`yV#Dw=;wOUs4#J%QnwijwQD8*;-^;7hp(*^&i8P_~+_ait zA#Pz+7ArmuAX{J}JK(-Vuvp3mDGCPG5h?!nUjXfy-Z$kujb12J78U2Au%KbA?qKqm zvXUvKWQu@;DT>jFF_O=CU7=Ah$(yl19N5PrO5$9z3auJPt=ho(u*~#F?3?wM3L;>E zI2|$F4)&$13cqH(rl-h9$A>6_+g=tY`~|cLj8<@WFNNrfhn?cEeQ8t)@4|$INmzJ( z_dXqjwDW?mtGjFlsH0wtBIsJ>Mx%{p&Pf$Q3buCAdxyUej{jf9{~hw9vj=~4(8COG z@fW_5cwj1~K|cgPUoQWKr<}b!|0!3K>bt#>*2Bm?3KLm78EZHUs~msQt)2Z*+LZSaaAlW)a=Sh!uL1Q{HRo{?ihmc!Yf z@t2u@d9i>j+<&RA=fw8^>*mV;q3qs2V=$9#$daXOmB(1Kj5S2a64^_ZNtP%~JQ!KV zZA8ivMOo)jc7?KJpFy6oR|iraJfH9K7kt03*Z2G5eVuci>pItU?(22# zbD#GcjD#$i6D9`&sgGkSEduV4ytrTL@N>`(e7wViWWsM43as;NIZ9?|k-Hg8*6a)6?aOQW2zNeuJ*$k6`cdz_Fv z6H3(udo2DEPxZr4Aclf}9#E+U6~Oqqg}*$ZHeh+BH+j?0n;c z%Z%!wcnb>i-l`@@@_?YwVGO)9Mk!-$q1?WE;8wYvsFF@ba|s?~~%@PHtFkt@Fk zc@2rd9?*GW1JgQ5;XQf|P>96)K-yuC1lxz2qPi2&4>iV5*9NkAaYe}Y%j9+I5qca%wAh{VHaxzQ(Vq~t7bo~aV89YK-!2E@7SU5}NL<5}~*8zLZ1 zbt2cnEP2~_yEx}T1z*BT?$OzqZ)r+CT}HRb^XqAGYZ~@a$-Z73*fB$Wo>IC`(1(T8 zrM)3mZ9jf?5P-d;DPrp{lucV#zv`c9-aZ*HpWb0|Q$>}7b)eaz5TLgA8uZH6CGcL3`27he1`bkDv5!YE89&%6&d~clG{? z{V8yIf{t~Sg4L1|iaJzW0`2Wf_ty$bJ}ed3nV!23%}0PEsvZ^gic;0wejKkmG2TrK z%V7m>HvK2qiw#smO$n9PWh_)s6|a9XHX06)4dVo3q*37i0DaY5BO@Y95sS}1JJG)~ zcS4%bjj8-OReVdXoOF@lyH}=l_R$R>KS<~K0iQQM)s>_*iIwZ!>0|6~3QMc1e_LGo z#Yf{uHlrc0_ybm2n$MY-Roo3DN%WF_z;W?U4<~7btl1Bf2psf5A;eR|oKs)k_&Stz zP(`#tz;z4BVv-v<`Xt|5kH;fK?0CC;R>Z1sAc%x@C+G=>o-ifKQB$~=1iudB0rQqR z0!kbazR9iOivfRQS@O-vB+Rr3AE0hD;JZ+0d44B*LRekPD^Cigiw;gGWZD${%eU^J z`d7$=m)V*KsB9)49JF1{#-me^ezvxybo>oG3Iue^M~p@#kwfyl$Yw#jU>gXSc+vra z5OWK1;J$TeO!jumBrFrdRLTCRPNgCMuMX3Wk1so~QTyGEG-h6gO+X5~) z1!2YFVd&wRMd2oDb8yb27tH4S1964VI#rS$ZOoo&GwcZ4Ri_8w)Y_=LuABFmC`A#_ z2fIdsHMU!Pni81OZHAGicJ!E#UyaM{{K{ki(d9Xv0(qI;o)8K1?|yIcr30v4FQNW2 zm8-t<#W1En_DzIpc?0N@qx?6^=O5E^viCR?nBlTw2kFhD^^RFvWsYBWAP#wNxKNp< z?Dxce1fobRv@hC8%?d;LuS;ZDB4FCNU)aMFE8eh*=e2stzqdosmjOb?SE<^t!Wy1& zN4JdH7|wr0f-1GQUp)KF!uXiZqR#@z${$LNpBv$-#Cz@};n`r?*&$%nX#sTI!(uQi z8q74;!8+RP5xZ$YUIl~GNe}OfI}k|Bl4ixjAz_0AlbNx%y5I+-Jf}qQG|0sr zfqzw{RF$`K(A@#TnxiypAWF!| zANW%`Qw7jO2$fFWCr7Tq3tuU4g;03oVDh{25i5 zxmHpvR;RZq)3!K+*!Az*lQZd3D5bmBxV2=C7P0o;)u`>~-PCfdDhoMdC zdp)gs*vU24Z~)h#4vCIHtWWXbwuLdh@Kc~*ivy4>i$bx`PKUq2VFP3To!_949ES<{ zvRxTQQwRs(;Y?D8KqAfnjc%roxaxz6f~Ilz;07R!X67;a)`D0#$qm1)@N(xRWOBuw z^K2%o(6=b?VZdo&w7!6c;ay=21Wg*_Cppz(6wNXMK_xvA#P`dxVSM;dr3Bo z8YYC~HYCt;_V`6Ecx5;z#y=A}so{c)Va)ISezb-EpNTeA~R z)w+o7GQ00Bpvdw?#nCj^$9uWAvE81W(AD?$&3{D|wZcuPr@Vf>byf7Oexs7>7~$2P zz$`0H)SG;oDaHZkkaeG2)ObArBu9Jv z1s6rk`=f*zTZ5E_sUi8DJm}=djE60wpB@n=uCm93vI0-+o)Bn6U)dURFhUBY20xD8 zVVw27v74VEaY#Sp!lNN=TvqPyP+W(;r?G}@B7HVYkKC^==8}|m@{U9DQyva^B|8ni z7NHL^;kV{aAQi@t;tBd77xSS2L~-Ii^yL7Ry%YnodSY#Q$-%zNVjUoGbx#lH10gg6 zgoW~XMPX@_@F0p++pvi@|FI=_G(sciA{M{fS#T9vVUFB;Tot16b3oI^9KlyLyEFzc&@FLgb!3Xuy6IQe7_F@9DH3G2*Oyl0KfiPg2DIm ztyiQ_eM^B+moRc&UUKxJ2H zXFW1RG-`M2xY$8*gVkQZ@%%8?FFH0x&&K5T{~03JogMVP_`#AB7x8&cL4E<@E87u{RDcT2iUaX>5pH&Uv-8#QPbt zksQ&2#tK|L;Shc9d~;}c(k<}bdqem73-TH%@{(~}XFdJ`3wZpI^h;OcMTm0qa)!i< z#ZM3Vnr9o$LmpI{$igX^6Q_EmTX<8;^R0rSPX8}=qCT7QY00seU=>3h~ zdF_j^4?4)fPzH0~PUM>kR>}MQgw z=3}#Fjs`O4jL0lOq+PtHyVl^7)$Qs_uwYy)eew6QySkFW@aB%x z+?%IcZ_2&7_rPjwGwmPAy)vuiXBiw)s^GL2T{?yi`>7IavjQ!n{7^7xzwVVT$<0NJ z;`e8;LTlDrb3@*8^uI0J@~z}YY=0vfV>}PKxeMx3$}#^mBp;6wY&aF#xJ=(DmPsuYV7oh8=Z8A2EL7&RBgghlL>lf1xxeZE$@2wY2bl*+qVvT z^II1WWCbAqZ^#G#O{bVND>@mbepFh~#k4PWJc*Jb5H(No@8;_=c?n|c;2k-6%?KxQNCsXJ07UkEL+xY+BhC~~!oSyPEjpFIiuG~=! zj!YZDK@3PHWM(Dgob51_-|XM`(A{rgY&XXwrP>`?BRF5yQ;3}?TekJOtTQ9kWPLP~ zlS6tj8!?R4P(^pqnO~^FNFDVvrA|IcgdR44uh&*GB{B8_L|<%wdd%>$eV@MzFzhdl zS~DVJnLkS?KeJ>m^Z~7Tu?MsLchqAY9mef*Jr`p1w~+J^(r{)0i6_Eg%n1#^a9 zN@HR@t^E_M>5ddtdUv^Mac(9^%Y8Ry`{!D0YEp1bQd#_m^ ZzCZj@c2{E!$VYsu7N#~AtBsNI{{jm%AM5}C diff --git a/docs/README.Rmd b/docs/README.Rmd index c6cfbed8..cbbb2d70 100644 --- a/docs/README.Rmd +++ b/docs/README.Rmd @@ -125,7 +125,7 @@ exp(25) Setting `k` to a smaller value results in less flexibility. This is because number of basis functions that contribute to functional behaviour is reduced ```{r, message=FALSE, warning=FALSE} lynx_mvgam_prior <- mvgam(data = lynx_train, - formula = population ~ s(season, bs = 'cc', k = 13), + formula = population ~ s(season, bs = 'cc', k = 12), knots = list(season = c(0.5, 19.5)), family = 'poisson', trend_model = 'AR3', @@ -143,7 +143,7 @@ In practice, imparting domain knowledge into prior specifications for penalised Users can also check what the default prior distributions are for given model formulations, which can be helpful to understand how the model can be modified but also to see any restrictions on what can be changed within the `mvgam` framework. ```{r} -test_priors <- get_mvgam_priors(population ~ s(season, bs = 'cc', k = 13), +test_priors <- get_mvgam_priors(population ~ s(season, bs = 'cc', k = 12), family = 'poisson', data = lynx_train, trend_model = 'AR3', @@ -155,7 +155,7 @@ Any of the above priors can be changed by modifying the `prior` column and suppl ```{r, message=FALSE, warning=FALSE} lynx_mvgam <- mvgam(data = lynx_train, newdata = lynx_test, - formula = population ~ s(season, bs = 'cc', k = 13), + formula = population ~ s(season, bs = 'cc', k = 12), knots = list(season = c(0.5, 19.5)), family = 'poisson', trend_model = 'AR3', diff --git a/docs/README.md b/docs/README.md index 41466089..7336bda4 100644 --- a/docs/README.md +++ b/docs/README.md @@ -218,7 +218,7 @@ behaviour is reduced ``` r lynx_mvgam_prior <- mvgam(data = lynx_train, - formula = population ~ s(season, bs = 'cc', k = 13), + formula = population ~ s(season, bs = 'cc', k = 12), knots = list(season = c(0.5, 19.5)), family = 'poisson', trend_model = 'AR3', @@ -255,7 +255,7 @@ be modified but also to see any restrictions on what can be changed within the `mvgam` framework. ``` r -test_priors <- get_mvgam_priors(population ~ s(season, bs = 'cc', k = 13), +test_priors <- get_mvgam_priors(population ~ s(season, bs = 'cc', k = 12), family = 'poisson', data = lynx_train, trend_model = 'AR3', @@ -268,11 +268,11 @@ test_priors #> 4 ar3 1 trend AR3 coefficient #> 5 sigma 1 trend sd #> prior example_change -#> 1 lambda ~ normal(10, 25); lambda ~ exponential(0.04); -#> 2 ar1 ~ std_normal(); ar1 ~ normal(-0.71, 0.45); -#> 3 ar2 ~ std_normal(); ar2 ~ normal(-0.29, 0.68); -#> 4 ar3 ~ std_normal(); ar3 ~ normal(0.82, 0.91); -#> 5 sigma ~ exponential(2); sigma ~ exponential(0.65); +#> 1 lambda ~ normal(10, 25); lambda ~ exponential(0.87); +#> 2 ar1 ~ std_normal(); ar1 ~ normal(0.67, 0.17); +#> 3 ar2 ~ std_normal(); ar2 ~ normal(0.67, 0.96); +#> 4 ar3 ~ std_normal(); ar3 ~ normal(-0.25, 0.17); +#> 5 sigma ~ exponential(2); sigma ~ exponential(0.74); ``` Any of the above priors can be changed by modifying the `prior` column @@ -286,7 +286,7 @@ and [here](https://mc-stan.org/cmdstanr/articles/cmdstanr.html)). ``` r lynx_mvgam <- mvgam(data = lynx_train, newdata = lynx_test, - formula = population ~ s(season, bs = 'cc', k = 13), + formula = population ~ s(season, bs = 'cc', k = 12), knots = list(season = c(0.5, 19.5)), family = 'poisson', trend_model = 'AR3', @@ -298,8 +298,8 @@ lynx_mvgam <- mvgam(data = lynx_train, #> Chain 3 Iteration: 1 / 1000 [ 0%] (Warmup) #> Chain 4 Iteration: 1 / 1000 [ 0%] (Warmup) #> Chain 4 Iteration: 100 / 1000 [ 10%] (Warmup) -#> Chain 3 Iteration: 100 / 1000 [ 10%] (Warmup) #> Chain 1 Iteration: 100 / 1000 [ 10%] (Warmup) +#> Chain 3 Iteration: 100 / 1000 [ 10%] (Warmup) #> Chain 2 Iteration: 100 / 1000 [ 10%] (Warmup) #> Chain 4 Iteration: 200 / 1000 [ 20%] (Warmup) #> Chain 1 Iteration: 200 / 1000 [ 20%] (Warmup) @@ -309,16 +309,16 @@ lynx_mvgam <- mvgam(data = lynx_train, #> Chain 1 Iteration: 300 / 1000 [ 30%] (Warmup) #> Chain 3 Iteration: 300 / 1000 [ 30%] (Warmup) #> Chain 2 Iteration: 300 / 1000 [ 30%] (Warmup) +#> Chain 4 Iteration: 400 / 1000 [ 40%] (Warmup) #> Chain 1 Iteration: 400 / 1000 [ 40%] (Warmup) #> Chain 3 Iteration: 400 / 1000 [ 40%] (Warmup) -#> Chain 4 Iteration: 400 / 1000 [ 40%] (Warmup) #> Chain 2 Iteration: 400 / 1000 [ 40%] (Warmup) +#> Chain 4 Iteration: 500 / 1000 [ 50%] (Warmup) +#> Chain 4 Iteration: 501 / 1000 [ 50%] (Sampling) #> Chain 3 Iteration: 500 / 1000 [ 50%] (Warmup) -#> Chain 3 Iteration: 501 / 1000 [ 50%] (Sampling) #> Chain 1 Iteration: 500 / 1000 [ 50%] (Warmup) -#> Chain 4 Iteration: 500 / 1000 [ 50%] (Warmup) +#> Chain 3 Iteration: 501 / 1000 [ 50%] (Sampling) #> Chain 1 Iteration: 501 / 1000 [ 50%] (Sampling) -#> Chain 4 Iteration: 501 / 1000 [ 50%] (Sampling) #> Chain 2 Iteration: 500 / 1000 [ 50%] (Warmup) #> Chain 2 Iteration: 501 / 1000 [ 50%] (Sampling) #> Chain 4 Iteration: 600 / 1000 [ 60%] (Sampling) @@ -327,28 +327,28 @@ lynx_mvgam <- mvgam(data = lynx_train, #> Chain 2 Iteration: 600 / 1000 [ 60%] (Sampling) #> Chain 4 Iteration: 700 / 1000 [ 70%] (Sampling) #> Chain 3 Iteration: 700 / 1000 [ 70%] (Sampling) -#> Chain 2 Iteration: 700 / 1000 [ 70%] (Sampling) #> Chain 1 Iteration: 700 / 1000 [ 70%] (Sampling) +#> Chain 2 Iteration: 700 / 1000 [ 70%] (Sampling) #> Chain 4 Iteration: 800 / 1000 [ 80%] (Sampling) #> Chain 3 Iteration: 800 / 1000 [ 80%] (Sampling) +#> Chain 1 Iteration: 800 / 1000 [ 80%] (Sampling) #> Chain 2 Iteration: 800 / 1000 [ 80%] (Sampling) #> Chain 4 Iteration: 900 / 1000 [ 90%] (Sampling) -#> Chain 1 Iteration: 800 / 1000 [ 80%] (Sampling) #> Chain 3 Iteration: 900 / 1000 [ 90%] (Sampling) #> Chain 2 Iteration: 900 / 1000 [ 90%] (Sampling) -#> Chain 4 Iteration: 1000 / 1000 [100%] (Sampling) -#> Chain 4 finished in 27.9 seconds. #> Chain 1 Iteration: 900 / 1000 [ 90%] (Sampling) +#> Chain 4 Iteration: 1000 / 1000 [100%] (Sampling) +#> Chain 4 finished in 28.8 seconds. #> Chain 3 Iteration: 1000 / 1000 [100%] (Sampling) -#> Chain 3 finished in 28.7 seconds. +#> Chain 3 finished in 28.9 seconds. #> Chain 2 Iteration: 1000 / 1000 [100%] (Sampling) -#> Chain 2 finished in 30.6 seconds. +#> Chain 2 finished in 31.3 seconds. #> Chain 1 Iteration: 1000 / 1000 [100%] (Sampling) -#> Chain 1 finished in 31.2 seconds. +#> Chain 1 finished in 31.7 seconds. #> #> All 4 chains finished successfully. -#> Mean chain execution time: 29.6 seconds. -#> Total execution time: 31.4 seconds. +#> Mean chain execution time: 30.2 seconds. +#> Total execution time: 31.8 seconds. ``` Inspect the resulting model file, which is written in the `Stan` @@ -368,7 +368,7 @@ code(lynx_mvgam) #> real p_coefs[1]; // prior locations for parametric coefficients #> matrix[total_obs, num_basis] X; // mgcv GAM design matrix #> int ytimes[n, n_series]; // time-ordered matrix (which col in X belongs to each [time, series] observation?) -#> matrix[11,11] S1; // mgcv smooth penalty matrix S1 +#> matrix[10,10] S1; // mgcv smooth penalty matrix S1 #> int n_nonmissing; // number of nonmissing observations #> int flat_ys[n_nonmissing]; // flattened nonmissing observations #> matrix[n_nonmissing, num_basis] flat_xs; // X values for nonmissing observations @@ -401,7 +401,7 @@ code(lynx_mvgam) #> b_raw[i] ~ normal(p_coefs[i], sqrt(1 / p_taus[i])); #> } #> // prior for s(season)... -#> b_raw[2:12] ~ multi_normal_prec(zero[2:12],S1[1:11,1:11] * lambda[1]); +#> b_raw[2:11] ~ multi_normal_prec(zero[2:11],S1[1:10,1:10] * lambda[1]); #> // priors for AR parameters #> ar1 ~ std_normal(); #> ar2 ~ std_normal(); @@ -515,7 +515,7 @@ effective sample sizes / mixing for all parameters ``` r summary(lynx_mvgam) #> GAM formula: -#> population ~ s(season, bs = "cc", k = 13) +#> population ~ s(season, bs = "cc", k = 12) #> #> Family: #> Poisson @@ -537,29 +537,28 @@ summary(lynx_mvgam) #> #> GAM coefficient (beta) estimates: #> 2.5% 50% 97.5% Rhat n.eff -#> (Intercept) 6.77727850 6.80142500 6.82540000 1.00 3385 -#> s(season).1 -0.72180295 -0.08288615 0.62589152 1.00 555 -#> s(season).2 -0.13235617 0.89959650 1.69452650 1.00 393 -#> s(season).3 0.03288385 1.53292500 2.53147750 1.01 298 -#> s(season).4 -0.14973782 0.94614900 1.88131050 1.01 424 -#> s(season).5 -1.04277900 -0.21305450 0.83163240 1.00 796 -#> s(season).6 -1.27960950 -0.27022850 0.88785662 1.01 398 -#> s(season).7 -0.87013257 0.08213570 1.08376250 1.00 748 -#> s(season).8 -0.61168777 0.71885650 1.75743225 1.01 349 -#> s(season).9 -0.90154072 0.47452100 1.78730050 1.01 254 -#> s(season).10 -1.17405325 -0.37126500 0.43564940 1.01 503 -#> s(season).11 -1.43829900 -0.81551100 -0.05862501 1.00 538 +#> (Intercept) 6.49464200 6.71836000 6.95850925 1.00 1798 +#> s(season).1 -0.57933102 0.05280635 0.69412617 1.01 729 +#> s(season).2 -0.17771063 0.83898350 1.78268900 1.01 335 +#> s(season).3 -0.06180266 1.24715000 2.45915425 1.01 313 +#> s(season).4 -0.46297547 0.43705650 1.33236650 1.00 690 +#> s(season).5 -1.11386325 -0.15558450 0.91460440 1.00 394 +#> s(season).6 -1.01934825 -0.02110400 1.05492275 1.00 544 +#> s(season).7 -0.61721910 0.37137650 1.35038950 1.00 575 +#> s(season).8 -0.99024972 0.24853500 1.73782475 1.01 279 +#> s(season).9 -1.18000250 -0.31756100 0.62468777 1.01 351 +#> s(season).10 -1.35713100 -0.69993050 -0.03220405 1.00 577 #> #> GAM smoothing parameter (rho) estimates: -#> 2.5% 50% 97.5% Rhat n.eff -#> s(season) 2.296071 3.400445 4.210571 1 493 +#> 2.5% 50% 97.5% Rhat n.eff +#> s(season) 2.082473 3.39916 4.244269 1 499 #> #> Latent trend parameter estimates: -#> 2.5% 50% 97.5% Rhat n.eff -#> ar1[1] 0.6836409 1.08386500 1.4045192 1.01 513 -#> ar2[1] -0.8218838 -0.40640350 0.0412204 1.00 1519 -#> ar3[1] -0.4356245 -0.06850925 0.3752050 1.01 400 -#> sigma[1] 0.3865736 0.47799150 0.6063589 1.00 1012 +#> 2.5% 50% 97.5% Rhat n.eff +#> ar1[1] 0.7075472 1.1172950 1.42644350 1 509 +#> ar2[1] -0.8062831 -0.3954970 0.03302338 1 1725 +#> ar3[1] -0.4639477 -0.1326695 0.27083892 1 480 +#> sigma[1] 0.3984874 0.4902760 0.63022597 1 968 #> #> Stan MCMC diagnostics #> n_eff / iter looks reasonable for all parameters @@ -643,7 +642,7 @@ the entire series (testing and training) ``` r plot(lynx_mvgam, type = 'forecast', newdata = lynx_test) #> Out of sample DRPS: -#> [1] 963.9811 +#> [1] 1078.426 #> ``` @@ -741,57 +740,57 @@ lynx_mvgam_poor <- mvgam(data = lynx_train, #> Chain 3 Iteration: 1 / 1000 [ 0%] (Warmup) #> Chain 4 Iteration: 1 / 1000 [ 0%] (Warmup) #> Chain 2 Iteration: 100 / 1000 [ 10%] (Warmup) -#> Chain 3 Iteration: 100 / 1000 [ 10%] (Warmup) +#> Chain 4 Iteration: 100 / 1000 [ 10%] (Warmup) #> Chain 1 Iteration: 100 / 1000 [ 10%] (Warmup) +#> Chain 3 Iteration: 100 / 1000 [ 10%] (Warmup) #> Chain 2 Iteration: 200 / 1000 [ 20%] (Warmup) +#> Chain 4 Iteration: 200 / 1000 [ 20%] (Warmup) #> Chain 1 Iteration: 200 / 1000 [ 20%] (Warmup) -#> Chain 2 Iteration: 300 / 1000 [ 30%] (Warmup) #> Chain 3 Iteration: 200 / 1000 [ 20%] (Warmup) -#> Chain 4 Iteration: 100 / 1000 [ 10%] (Warmup) +#> Chain 2 Iteration: 300 / 1000 [ 30%] (Warmup) +#> Chain 4 Iteration: 300 / 1000 [ 30%] (Warmup) +#> Chain 3 Iteration: 300 / 1000 [ 30%] (Warmup) #> Chain 1 Iteration: 300 / 1000 [ 30%] (Warmup) +#> Chain 4 Iteration: 400 / 1000 [ 40%] (Warmup) #> Chain 2 Iteration: 400 / 1000 [ 40%] (Warmup) -#> Chain 3 Iteration: 300 / 1000 [ 30%] (Warmup) -#> Chain 4 Iteration: 200 / 1000 [ 20%] (Warmup) +#> Chain 3 Iteration: 400 / 1000 [ 40%] (Warmup) #> Chain 1 Iteration: 400 / 1000 [ 40%] (Warmup) +#> Chain 4 Iteration: 500 / 1000 [ 50%] (Warmup) #> Chain 2 Iteration: 500 / 1000 [ 50%] (Warmup) -#> Chain 3 Iteration: 400 / 1000 [ 40%] (Warmup) -#> Chain 4 Iteration: 300 / 1000 [ 30%] (Warmup) +#> Chain 4 Iteration: 501 / 1000 [ 50%] (Sampling) #> Chain 2 Iteration: 501 / 1000 [ 50%] (Sampling) #> Chain 1 Iteration: 500 / 1000 [ 50%] (Warmup) -#> Chain 2 Iteration: 600 / 1000 [ 60%] (Sampling) #> Chain 3 Iteration: 500 / 1000 [ 50%] (Warmup) -#> Chain 4 Iteration: 400 / 1000 [ 40%] (Warmup) #> Chain 1 Iteration: 501 / 1000 [ 50%] (Sampling) #> Chain 3 Iteration: 501 / 1000 [ 50%] (Sampling) -#> Chain 3 Iteration: 600 / 1000 [ 60%] (Sampling) -#> Chain 4 Iteration: 500 / 1000 [ 50%] (Warmup) +#> Chain 4 Iteration: 600 / 1000 [ 60%] (Sampling) +#> Chain 2 Iteration: 600 / 1000 [ 60%] (Sampling) #> Chain 1 Iteration: 600 / 1000 [ 60%] (Sampling) +#> Chain 3 Iteration: 600 / 1000 [ 60%] (Sampling) +#> Chain 4 Iteration: 700 / 1000 [ 70%] (Sampling) #> Chain 2 Iteration: 700 / 1000 [ 70%] (Sampling) -#> Chain 4 Iteration: 501 / 1000 [ 50%] (Sampling) -#> Chain 3 Iteration: 700 / 1000 [ 70%] (Sampling) -#> Chain 2 Iteration: 800 / 1000 [ 80%] (Sampling) -#> Chain 4 Iteration: 600 / 1000 [ 60%] (Sampling) #> Chain 1 Iteration: 700 / 1000 [ 70%] (Sampling) -#> Chain 3 Iteration: 800 / 1000 [ 80%] (Sampling) -#> Chain 2 Iteration: 900 / 1000 [ 90%] (Sampling) -#> Chain 4 Iteration: 700 / 1000 [ 70%] (Sampling) -#> Chain 1 Iteration: 800 / 1000 [ 80%] (Sampling) -#> Chain 2 Iteration: 1000 / 1000 [100%] (Sampling) -#> Chain 3 Iteration: 900 / 1000 [ 90%] (Sampling) -#> Chain 2 finished in 3.6 seconds. #> Chain 4 Iteration: 800 / 1000 [ 80%] (Sampling) -#> Chain 3 Iteration: 1000 / 1000 [100%] (Sampling) -#> Chain 3 finished in 3.8 seconds. -#> Chain 1 Iteration: 900 / 1000 [ 90%] (Sampling) +#> Chain 2 Iteration: 800 / 1000 [ 80%] (Sampling) +#> Chain 1 Iteration: 800 / 1000 [ 80%] (Sampling) +#> Chain 3 Iteration: 700 / 1000 [ 70%] (Sampling) #> Chain 4 Iteration: 900 / 1000 [ 90%] (Sampling) +#> Chain 1 Iteration: 900 / 1000 [ 90%] (Sampling) +#> Chain 2 Iteration: 900 / 1000 [ 90%] (Sampling) +#> Chain 3 Iteration: 800 / 1000 [ 80%] (Sampling) #> Chain 1 Iteration: 1000 / 1000 [100%] (Sampling) #> Chain 4 Iteration: 1000 / 1000 [100%] (Sampling) -#> Chain 1 finished in 4.2 seconds. -#> Chain 4 finished in 4.2 seconds. +#> Chain 1 finished in 7.7 seconds. +#> Chain 4 finished in 7.7 seconds. +#> Chain 2 Iteration: 1000 / 1000 [100%] (Sampling) +#> Chain 2 finished in 7.8 seconds. +#> Chain 3 Iteration: 900 / 1000 [ 90%] (Sampling) +#> Chain 3 Iteration: 1000 / 1000 [100%] (Sampling) +#> Chain 3 finished in 9.0 seconds. #> #> All 4 chains finished successfully. -#> Mean chain execution time: 3.9 seconds. -#> Total execution time: 4.3 seconds. +#> Mean chain execution time: 8.0 seconds. +#> Total execution time: 9.1 seconds. ``` We choose a set of timepoints within the training data to forecast from, @@ -813,10 +812,10 @@ markedly better (far lower DRPS) for this evaluation timepoint ``` r summary(mod1_eval$series1$drps) #> Min. 1st Qu. Median Mean 3rd Qu. Max. -#> 30.18 67.19 191.25 167.38 236.59 323.62 +#> 58.94 91.37 162.65 164.15 188.98 320.40 summary(mod2_eval$series1$drps) #> Min. 1st Qu. Median Mean 3rd Qu. Max. -#> 33.23 38.45 313.87 289.71 465.96 675.85 +#> 54.24 60.54 266.41 265.25 398.39 624.62 ``` Nominal coverages for both models’ 90% prediction intervals @@ -825,7 +824,7 @@ Nominal coverages for both models’ 90% prediction intervals mean(mod1_eval$series1$in_interval) #> [1] 0.9 mean(mod2_eval$series1$in_interval) -#> [1] 0.7 +#> [1] 0.8 ``` The `compare_mvgams` function automates this process by rolling along a diff --git a/man/get_monitor_pars.Rd b/man/get_monitor_pars.Rd index 0fe5afad..197a2d38 100644 --- a/man/get_monitor_pars.Rd +++ b/man/get_monitor_pars.Rd @@ -7,7 +7,7 @@ get_monitor_pars(family, smooths_included = TRUE, use_lv, trend_model, drift) } \arguments{ -\item{family}{\code{character}. Must be either 'nb' (for Negative Binomial), 'tw' (for Tweedie) or 'poisson'} +\item{family}{\code{character}} \item{smooths_included}{Logical. Are smooth terms included in the model formula?} diff --git a/manuscript_analysis/Case studies/mvgam_case_study1.Rmd b/manuscript_analysis/Case studies/mvgam_case_study1.Rmd index 68f905c2..46756492 100644 --- a/manuscript_analysis/Case studies/mvgam_case_study1.Rmd +++ b/manuscript_analysis/Case studies/mvgam_case_study1.Rmd @@ -2,10 +2,10 @@ title: 'mvgam case study 1: model comparison and data assimilation' author: "Nicholas Clark (n.clark@uq.edu.au)" output: - pdf_document: - highlight: zenburn html_document: df_print: paged + pdf_document: + highlight: zenburn word_document: default urlcolor: blue --- @@ -372,7 +372,7 @@ In fact the the autocorrelated observation model is overestimating the amount of data. This can partially explain the exponentially growing uncertainties in the mgcv forecast The truth for the ovderdispersion parameter is `10`. This is likely driven by the autoregressive model's high sensitivity to the noisy observation process. Observations in ecology very often display properties of overdispersion, and this results in observation 'outliers' that can have large leverage on resulting parameter estimates for autoregressive models. The noisy observation process is not a problem for latent trend models such as those estimated by `mvgam`, as the observation model is estimated separately from the temporal process (as is the norm for state-space models) ```{r} m_mgcv$family$getTheta(TRUE) -quantile(MCMCvis::MCMCchains(m_mvgam$model_output, 'r'), c(0.1, 0.5, 0.9)) +quantile(MCMCvis::MCMCchains(m_mvgam$model_output, 'phi'), c(0.1, 0.5, 0.9)) ``` Looking at estimates for the full trend and forecast distributions shows how the `mvgam` model is clearly doing a decent job of tracking the latent state up to the end of the training period while also giving reasonable estimates for the overdispersion in the data ```{r} @@ -842,9 +842,8 @@ lynx_full = dplyr::mutate(lynx_full, lag6 = dplyr::lag(popl,6, default = mean_pop_l)) ``` -For `mvgam` models, the response needs to be labelled `y` and we also need an indicator of the series name as a `factor` variable +For `mvgam` models, we need an indicator of the series name as a `factor` variable ```{r} -lynx_full$y <- lynx_full$population lynx_full$series <- factor('series1') ``` @@ -1044,10 +1043,10 @@ This forecast is highly overconfident, with very unrealistic uncertainty interva ```{r, message=FALSE, warning=FALSE} lynx_mvgam <- mvgam(data = lynx_train, newdata = lynx_test, - formula = y ~ s(season, bs = 'cc', k = 19), + formula = population ~ s(season, bs = 'cc', k = 15), knots = list(season = c(0.5, 19.5)), family = 'poisson', - trend_model = 'AR1', + trend_model = 'AR2', use_stan = TRUE, chains = 4) ``` diff --git a/manuscript_analysis/Case studies/mvgam_case_study1.html b/manuscript_analysis/Case studies/mvgam_case_study1.html index 9ceb0caa..5a541f07 100644 --- a/manuscript_analysis/Case studies/mvgam_case_study1.html +++ b/manuscript_analysis/Case studies/mvgam_case_study1.html @@ -1916,22 +1916,22 @@

Nicholas Clark (Nicholas Clark (Nicholas Clark (Nicholas Clark (Nicholas Clark (Nicholas Clark (mod_skeleton$inits 
## function() {
 ##               list(b_raw = runif(model_data$num_basis, -2, 2),
-##                    r_inv = runif(NCOL(model_data$ytimes), 1, 50),
+##                    phi_inv = runif(NCOL(model_data$ytimes), 1, 50),
 ##                    sigma = runif(model_data$n_series, 0.075, 1))
 ##             }
-## <bytecode: 0x0000011f336b0530>
-## <environment: 0x0000011f2f80ad08>
+## <bytecode: 0x00000204dad10fc0> +## <environment: 0x00000204d94a2c08> 
head(mod_skeleton$pregam)
## $m
 ## [1] 1
@@ -2233,7 +2233,7 @@ 
Nicholas Clark (Nicholas Clark (Nicholas Clark (Nicholas Clark (Nicholas Clark (Nicholas Clark (Nicholas Clark (

+

 
Generally the uncertainy blows out when forecasting further ahead for
 the mgcv model (why might this be?); if we only look 2
 timesteps ahead, it is not quite as terrible but it becomes more clear
@@ -2441,15 +2441,15 @@ 
Nicholas Clark (

+

 
Why does autocorrelated observation model perform so badly? Q-Q plots
 give some insight, as they clearly indicate the dynamic latent trend
 model shows a better fit of the observation process to the data compared
 to the autocorrelated observation model

 
plot(m_mvgam, 'residuals')

-

+

 
qq.gam(m_mgcv, pch = 16, lwd = 3, cex = 0.8)

 

 
In fact the the autocorrelated observation model is overestimating
@@ -2466,9 +2466,9 @@ 
Nicholas Clark (m_mgcv$family$getTheta(TRUE)
## [1] 2.167732
-
quantile(MCMCvis::MCMCchains(m_mvgam$model_output, 'r'), c(0.1, 0.5, 0.9))
+
quantile(MCMCvis::MCMCchains(m_mvgam$model_output, 'phi'), c(0.1, 0.5, 0.9))
##       10%       50%       90% 
-##  4.359946 11.529510 91.821257
+## 3.953412 9.800959 55.961303

Looking at estimates for the full trend and forecast distributions shows how the mvgam model is clearly doing a decent job of tracking the latent state up to the end of the training period while @@ -2478,13 +2478,13 @@

Nicholas Clark (

+

layout(1)
plot(m_mvgam, 'forecast', newdata = data_test)
## Out of sample DRPS:
-
## [1] 63.12883
+
## [1] 61.0694
##
-

+

We can get a better idea of what is happening for the autoregressive observation model by plotting posterior realisations on the log scale. Here each line is a realisation, and we show these lines as a spaghetti @@ -2522,7 +2522,7 @@

Nicholas Clark (

+

layout(1)

From the plots above it is apparent that although some of the realisations are sensible, as soon as a path climbs upward it tends to @@ -2568,9 +2568,9 @@

Nicholas Clark (

+

We should also re-check the autoregressive model’s realisations, which look much better now

plot(1, type = "n", bty = 'L',
@@ -2737,66 +2737,66 @@ 
Nicholas Clark (plot(m_mvgam, 'smooths')
-

+

Inspect posterior realisations of the smooth function

plot(m_mvgam, 'smooths', realisations = TRUE)
-

+

The plot_mvgam_smooth function allows more flexibility for plotting smooth functions, including an ability to supply newdata for plotting posterior marginal simulations. @@ -2807,29 +2807,29 @@

Nicholas Clark (plot_mvgam_smooth(m_mvgam, series = 1, smooth = 'p_whiting', residuals = TRUE) -

+

We can also perform a series of posterior predictive checks (using the ppc() function) to see if the model is able to simulate data for the training period that looks realistic and unbiased. First, examine simulated histograms for posterior predictions (yhat) and compare to the observations (y)

ppc(m_mvgam, series = 1, type = 'hist', legend_position = 'bottomright')
-

+

Now plot the distribution of predicted means compared to the observed mean

ppc(m_mvgam, series = 1, type = 'mean')
-

+

Next examine simulated empirical Cumulative Distribution Functions (CDF) for posterior predictions (yhat) and compare to the CDF of the observations (y)

ppc(m_mvgam, series = 1, type = 'cdf')
-

+

Finally look for any biases in predictions by examining a Probability Integral Transform (PIT) histogram. If our predictions are not biased one way or another (i.e. not consistently under- or over-predicting), this histogram should look roughly uniform

ppc(m_mvgam, series = 1, type = 'pit')
-

+

All of these plots indicate the model is well calibrated against the training data, with no apparent pathological behaviors exhibited. Now for some investigation of the estimated relationships and forecasts. We @@ -2838,7 +2838,7 @@

Nicholas Clark (plot(m_mvgam, series = 1, type = 'residuals') -

+

Ok so the model is doing well when fitting against the training data, but how are its forecasts? We can now evaluate the forecasts of the two competing models just as we did above for the simulated data

@@ -2852,9 +2852,9 @@

Nicholas Clark (

+

Both models fail to anticipate the upward swing in landings following the end of the training data, but as with the simulations above, the autocorrelated observation model fitted by mgcv is @@ -2899,9 +2899,9 @@

Nicholas Clark (

+

This model shows an improvement in forecasting ability as the uncertainty in the temporal smooth function is more realistically growing into the future. But again, the dynamic GAM is superior in @@ -2937,9 +2937,9 @@

Nicholas Clark (

+

The dynamic GAM is difficult to beat for this example. But what other utilities does the mvgam package make available? We can also re-do the posterior predictive checks, but this time focusing only @@ -2947,11 +2947,11 @@

Nicholas Clark (ppc(m_mvgam, series = 1, type = 'mean', newdata = data_test) -

+

ppc(m_mvgam, series = 1, type = 'cdf', newdata = data_test)
-

+

ppc(m_mvgam, series = 1, type = 'pit', newdata = data_test)
-

+

There are some problems with the way this model is generating future predictions, so we would need to perform a more rigorous and principled model development strategy to improve our model’s predictive abilities. @@ -2978,71 +2978,71 @@

Nicholas Clark (plot(m_mvgam2, series = 1, type = 'forecast', newdata = data_test) 
## Out of sample DRPS:
-
## [1] 2005.684
+
## [1] 2029.195
##
-

+

plot(m_mvgam, series = 1, type = 'forecast', newdata = data_test)
## Out of sample DRPS:
-
## [1] 1841.884
+
## [1] 1867.568
##
-

+

Now we will showcase how different dynamic models can be compared using rolling probabilistic forecast evaluation, which is especially useful if we don’t already have out of sample observations for comparing @@ -3065,14 +3065,14 @@

Nicholas Clark (compare_mvgams(model1 = m_mvgam, model2 = m_mvgam2, fc_horizon = 6, n_evaluations = 30, n_cores = 3) 
## DRPS summaries per model (lower is better)
-##             Min.  1st Qu.    Median      Mean   3rd Qu.      Max.
-## Model 1 1391.280 2246.340  3749.271  3666.239  4881.474  6103.371
-## Model 2 8982.565 9979.114 11715.380 11654.366 12957.589 14740.851
+##             Min.   1st Qu.    Median      Mean  3rd Qu.      Max.
+## Model 1 1393.951  2254.688  3755.298  3677.073  4865.27  6168.148
+## Model 2 9137.359 10147.497 11904.739 11848.968 13190.58 14966.034
 ## 
 ## 90% interval coverages per model (closer to 0.9 is better)
-## Model 1 0.8833333 
+## Model 1 0.8777778 
 ## Model 2 0.4722222
-

+

The series of plots generated by compare_mvgams clearly show that the first dynamic model generates better predictions. In each plot, DRPS for the forecast horizon is lower for the first @@ -3100,7 +3100,7 @@

Nicholas Clark (pfilter_mvgam_init(object = m_mvgam, n_particles = 20000, n_cores = 3, data_assim = model_dat[28,]) 
## Saving particles to pfilter/particles.rda 
-##  ESS = 2459.189
+## ESS = 2330.044

Now we are ready to run the particle filter. This function will assimilate the next two out of sample observations in data_test and update the forecast after each assimilation @@ -3119,12 +3119,12 @@

Nicholas Clark (

+

Here it is apparent that the distribution has shifted upward in light of the 3 observations that have been assimilated. This is an advantageous way of allowing a model to slowly adapt to new @@ -3215,11 +3215,9 @@

Nicholas Clark (lynx_full$y <- lynx_full$population -lynx_full$series <- factor('series1') +

For mvgam models, we need an indicator of the series +name as a factor variable

+
lynx_full$series <- factor('series1')

Split the data into training (first 40 years) and testing (next 10 years of data) to evaluate multi-step ahead forecasts

lynx_train = lynx_full[1:40, ]
@@ -3473,10 +3471,10 @@ 
Nicholas Clark (lynx_mvgam <- mvgam(data = lynx_train,
                newdata = lynx_test,
-               formula = y ~ s(season, bs = 'cc', k = 19),
+               formula = population ~ s(season, bs = 'cc', k = 15),
                knots = list(season = c(0.5, 19.5)),
                family = 'poisson',
-               trend_model = 'AR1',
+               trend_model = 'AR2',
                use_stan = TRUE,
                chains = 4)
## Running MCMC with 4 parallel chains...
@@ -3485,58 +3483,58 @@ 
Nicholas Clark (fits <- MCMCvis::MCMCchains(lynx_mvgam$model_output, 'ypred')
@@ -3559,7 +3557,7 @@ 
Nicholas Clark (

+

 
Calculate DRPS over the 10-year horizon for the mvgam model

 
lynx_mvgam_drps <- drps_mcmc_object(truth = lynx_test$population[1:10], 
                    fc = fits)

@@ -3579,7 +3577,7 @@ 
Nicholas Clark (sum(lynx_mvgam_drps[,1])
-
## [1] 808.6506
+
## [1] 1046.792
mean(lynx_mvgam_drps[,2])
## [1] 1

The mvgam has much more realistic uncertainty than the @@ -3592,7 +3590,7 @@

Nicholas Clark (summary(lynx_mvgam) 
## GAM formula:
-
## y ~ s(season, bs = "cc", k = 19)
+
## population ~ s(season, bs = "cc", k = 15)
## 
## Family:
## Poisson
@@ -3601,7 +3599,7 @@

Nicholas Clark (Nicholas Clark (Nicholas Clark (plot(lynx_mgcv, select=1, shade=T) 

plot_mvgam_smooth(lynx_mvgam, 1, 'season')
-

+

We can also view the mvgam’s posterior predictions for the entire series (testing and training)

plot(lynx_mvgam, type = 'forecast', newdata = lynx_test)
## Out of sample DRPS:
-
## [1] 808.6506
+
## [1] 1046.792
##
-

+

And the estimated trend

plot(lynx_mvgam, type = 'trend', newdata = lynx_test)
-

+

A key aspect of ecological forecasting is to understand how different components of a model contribute to forecast uncertainty. We can estimate contributions to forecast uncertainty for the GAM smooth functions and the latent trend using mvgam

plot(lynx_mvgam, type = 'uncertainty', newdata = lynx_test)
-

+

Both components contribute to forecast uncertainty, with the trend component contributing more over time (as it should since this is the stochastic forecast component). This suggests we would still need some @@ -3694,7 +3689,7 @@

Nicholas Clark (plot_mvgam_resids(lynx_mvgam, n_bins = 25) -

+

References
Clark, N.J. and Wells, K. (2022). Dynamic Generalized Additive Models (DGAMs) for forecasting discrete ecological time series. Methods in diff --git a/manuscript_analysis/Case studies/mvgam_case_study3.Rmd b/manuscript_analysis/Case studies/mvgam_case_study3.Rmd index 40eef55d..8384c88b 100644 --- a/manuscript_analysis/Case studies/mvgam_case_study3.Rmd +++ b/manuscript_analysis/Case studies/mvgam_case_study3.Rmd @@ -2,10 +2,10 @@ title: 'mvgam case study 3: distributed lag models' author: "Nicholas Clark (n.clark@uq.edu.au)" output: - pdf_document: - highlight: zenburn html_document: df_print: paged + pdf_document: + highlight: zenburn word_document: default urlcolor: blue --- @@ -112,7 +112,7 @@ data_test <- list(lag = data_all$lag[175:length(data_all$y),], mintemp = data_all$mintemp[175:length(data_all$y),]) ``` -Now we can fit a Bayesian GAM with distributed lag terms for precipitation and minimum temperature. The distributed lags are set up as tensor product smooth functions (see `help(te)` for an explanation of tensor product smooth constructions in the `mgcv` package) between `lag` and each covariate. We will start simply by assuming our data follow a `Poisson` observation process +Now we can fit a Bayesian GAM with distributed lag terms for precipitation and minimum temperature. The distributed lags are set up as tensor product smooth functions (see `help(te)` for an explanation of tensor product smooth constructions in the `mgcv` package) between `lag` and each covariate. We will start simply by assuming our data follow a `Poisson` observation process. Note that the Gibbs samplers used by `JAGS` are not efficient at exploring high-dimensional posteriors, so we cannot expect reasonable performance without running the chains for many iterations. For this example, I only use a burnin of `5000` ```{r, message=FALSE, warning=FALSE} mod1 <- mvgam(formula = y ~ te(mintemp, lag, k = c(6, 6)) + te(precip, lag, k = c(6, 6)), @@ -120,7 +120,7 @@ mod1 <- mvgam(formula = y ~ te(mintemp, lag, k = c(6, 6)) + newdata = data_test, family = 'poisson', chains = 4, - burnin = 15000, + burnin = 5000, trend_model = 'None') ``` @@ -143,7 +143,7 @@ mod2 <- mvgam(formula = y ~ te(mintemp, lag, k = c(6, 6)) + newdata = data_test, family = 'tw', chains = 4, - burnin = 15000, + burnin = 5000, trend_model = 'None') ``` @@ -181,16 +181,16 @@ mod3 <- mvgam(formula = y ~ te(mintemp, lag, k = c(6, 6)) + family = 'tw', priors = priors, chains = 4, - burnin = 15000, + burnin = 5000, trend_model = 'AR3') summary(mod3) ``` A pairs plot of the logged versions of the latent trend precision and the overdispersion parameter can give insight into whether there is still any strange behaviour or problematic parameter spaces that we must think more deeply about ```{r} -plot(log(MCMCvis::MCMCchains(mod3$model_output, 'twdis')), +plot(log(MCMCvis::MCMCchains(mod3$model_output, 'phi')), log(MCMCvis::MCMCchains(mod3$model_output, 'tau')), - ylab = 'log(tau)', xlab = 'log(twdis)', pch = 16, col = "#8F272740") + ylab = 'log(tau)', xlab = 'log(phi)', pch = 16, col = "#8F272740") ``` Our rootogram has not improved much with the addition of the latent trend diff --git a/manuscript_analysis/Case studies/mvgam_case_study3.html b/manuscript_analysis/Case studies/mvgam_case_study3.html index ffc381a3..699dfdb3 100644 --- a/manuscript_analysis/Case studies/mvgam_case_study3.html +++ b/manuscript_analysis/Case studies/mvgam_case_study3.html @@ -1737,7 +1737,7 @@

Nicholas Clark (plot_mvgam_series(data = data_all) -

+

Create training and testing sets; start at observation 7 so that the NA values at the beginning of the covariate lag matrices are not included. Currently there is no option for on-the-fly imputation @@ -1766,14 +1766,18 @@

Nicholas Clark (

ppc(mod1, type = 'rootogram') -

+

The Poisson model is not doing a great job of capturing dispersion, underpredicting the zeros in the data and overpredicting some of the medium-range values (counts of ~5-30). The residual Q-Q plot confirms that the Poisson is not an appropriate distribution for these data

plot(mod1, type = 'residuals')
-

+

Given the overdispersion present in the data, we will now assume a Geometric-Poisson observation model, which can be more flexible than the Negative binomial for modelling @@ -1807,15 +1811,15 @@

Nicholas Clark (ppc(mod2, type = 'rootogram') -

+

However, the residual plot looks much better for this model

plot(mod2, type = 'residuals')
-

+

The summary of the model provides useful information on convergence for unobserved parameters. Notice how strongly positive the overdispersion parameter is estimated to be, providing further evidence @@ -1847,89 +1851,89 @@

Nicholas Clark (Nicholas Clark (