Evaluate performance and accuracy of summation algorithms

- sum_naive: naive summation, slowest, bad accuracy
- sum_pairwise_recur: recursive pairwise sum, good accuracy
- sum_pairwise_itera: non-recursive parirwise sum, fastest, good accuracy

Apple M1
--------------------------------------------------------------------
Benchmark                          Time             CPU   Iterations
--------------------------------------------------------------------
BM_naive/4096                   3771 ns         3771 ns       183780
BM_naive/65536                 61464 ns        61464 ns        11302
BM_naive/1048576              986057 ns       986058 ns          706
BM_pairwise_recur/4096          2050 ns         2050 ns       340720
BM_pairwise_recur/65536        32807 ns        32807 ns        21337
BM_pairwise_recur/1048576     537114 ns       537119 ns         1272
BM_pairwise_itera/4096           903 ns          903 ns       770001
BM_pairwise_itera/65536        13817 ns        13818 ns        50421
BM_pairwise_itera/1048576     227697 ns       227697 ns         3062

Ryzen 5 4500U
--------------------------------------------------------------------
Benchmark                          Time             CPU   Iterations
--------------------------------------------------------------------
BM_naive/4096                   3190 ns         3190 ns       219190
BM_naive/65536                 52609 ns        52609 ns        11968
BM_naive/1048576              878888 ns       878888 ns          794
BM_pairwise_recur/4096          1711 ns         1711 ns       413176
BM_pairwise_recur/65536        26952 ns        26952 ns        27751
BM_pairwise_recur/1048576     590021 ns       590022 ns         1147
BM_pairwise_itera/4096          1072 ns         1072 ns       652425
BM_pairwise_itera/65536        17174 ns        17174 ns        41780
BM_pairwise_itera/1048576     505832 ns       505833 ns         1256

Xeon Gold 5218
-----------------------------------------------------------------
Benchmark                          Time           CPU Iterations
-----------------------------------------------------------------
BM_naive/4096                   7060 ns       7060 ns      99137
BM_naive/65536                114232 ns     114234 ns       6127
BM_naive/1048576             1831420 ns    1831404 ns        382
BM_pairwise_recur/4096          2496 ns       2497 ns     280879
BM_pairwise_recur/65536        41360 ns      41360 ns      16926
BM_pairwise_recur/1048576     681734 ns     681730 ns       1027
BM_pairwise_itera/4096          2170 ns       2170 ns     322486
BM_pairwise_itera/65536        34210 ns      34211 ns      20458
BM_pairwise_itera/1048576     565401 ns     565414 ns       1237