Wilcoxon refactor

[Intron7]

Implement a better GPU accelerated version of wilcoxon in rapids-singlecell inspired by the super fast illico verison.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 81.24031% with 121 lines in your changes missing coverage. Please review.
✅ Project coverage is 87.41%. Comparing base (c785fbb) to head (f8c00d8).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
...s_singlecell/tools/_rank_genes_groups/_wilcoxon.py	79.12%	91 Missing ⚠️
...apids_singlecell/tools/_rank_genes_groups/_core.py	80.88%	26 Missing ⚠️
...pids_singlecell/tools/_rank_genes_groups/_utils.py	87.87%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #636      +/-   ##
==========================================
- Coverage   88.91%   87.41%   -1.51%     
==========================================
  Files         103      103              
  Lines        8464     8953     +489     
==========================================
+ Hits         7526     7826     +300     
- Misses        938     1127     +189     

Files with missing lines	Coverage Δ
...ds_singlecell/tools/_rank_genes_groups/__init__.py	87.95% <100.00%> (+5.49%)	⬆️
...ecell/tools/_rank_genes_groups/_wilcoxon_binned.py	83.33% <100.00%> (-0.24%)	⬇️
...pids_singlecell/tools/_rank_genes_groups/_utils.py	68.04% <87.87%> (-17.88%)	⬇️
...apids_singlecell/tools/_rank_genes_groups/_core.py	73.66% <80.88%> (-18.07%)	⬇️
...s_singlecell/tools/_rank_genes_groups/_wilcoxon.py	80.63% <79.12%> (-16.99%)	⬇️

... and 1 file with indirect coverage changes