feat: add silv_predict_biomass_auto and silv_predict_biomass_components

[aitorvv]

This PR implements two new features for predictive biomass modelling:

1. silv_predict_biomass_auto(): automatically selects the best available model per species/component based on priority tables, with automatic fallback to montero-2005 (which doesn't require height) if tree height is NULL, NA, or 0.
2. silv_predict_biomass_components(): returns a wide-format data frame containing predicted weights for all pure structural components (stem, bark, branches, roots, etc.) for each tree. It also implements AGB/BGB capitalization consistency and species support validation.
3. Bugfix in core silv_predict_biomass() to prevent returning empty lists/NULLs when height = NULL is supplied.

Verification

# Inspect new/modified datasets
devtools::load_all()

# 1. Verify auto priority model selection (Pinus pinaster with height -> ruiz-peinado-2011)
silv_predict_biomass_auto("Pinus pinaster", 20, 12, component = "tree")

# 2. Verify height fallback (height is NULL -> forces montero-2005)
silv_predict_biomass_auto("Pinus pinaster", 20, NULL, component = "tree")

# 3. Verify component calculation with uppercase column names
silv_predict_biomass_components(
  species = "Fagus sylvatica",
  diameter = 2,
  height = 3,
  model_fn = eq_biomass_menendez_2022,
  rcd = 2
)

# 4. Verify early abort for unsupported species
try(silv_predict_biomass_components(
  species = "Quercus petraea",
  diameter = 25,
  height = 15,
  model_fn = eq_biomass_ruiz_peinado_2012
))