[python] Optimization of ExperimentAxisQuery to_anndata

[bkmartinjr]

Apply partitioning optimization and improved concurrency to the ExperimentAxisQuery.to_anndata, similar to the approach used in #3328

Other changes related to this:

• removed numba package dependency, as it is no longer needed
• removal of dead code in both ExperimentAxisQuery and the old CSR converter
• remove obsolete comments related to resource handling (no more reference cycles now that we use the context thread pool).
• fixed a constant value typo in outgest
• style/cleanup of small amount of code in ExperimentAxisQuery._read
• improve empty matrix handling in CompresedMatrix.from_soma

As part of testing I validated this on both the sparse-with-dups and sparse-without-dups readers via S3. There was a performance improvement on both.

[codecov]

Codecov Report

Attention: Patch coverage is 94.33962% with 3 lines in your changes missing coverage. Please review.

Project coverage is 85.85%. Comparing base (ea3cd1a) to head (155fcad).
Report is 4 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3359      +/-   ##
==========================================
+ Coverage   85.64%   85.85%   +0.20%     
==========================================
  Files          57       56       -1     
  Lines        6201     6107      -94     
==========================================
- Hits         5311     5243      -68     
+ Misses        890      864      -26     

Flag	Coverage Δ
python	85.85% <94.33%> (+0.20%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Components	Coverage Δ
python_api	85.85% <94.33%> (+0.20%)	⬆️
libtiledbsoma	∅ <ø> (∅)

---- 🚨 Try these New Features:

• Flaky Tests Detection - Detect and resolve failed and flaky tests

[johnkerl]

This is super-cool @bkmartinjr ! Just some minor comments.

[johnkerl]

🚢
Thanks @bkmartinjr !! :)

[johnkerl]

Thanks @bkmartinjr !