[ENH] Efficient out-of-bag computation per honest tree

[adam2392]

Right now, in order to have "out-of-bag" samples in the sklearn pipeline, one needs to turn bootstrap=True for the forest. The sample_weights are adjusted accordingly to "bootstrap" the samples.

This PR will allow one to access the OOB sample indices from the forest so MIGHT can take advantage of those for estimating the population statistics (e.g. S@S98, or MI).

Another issue is how can we make out-of-bag samples without bootstrap? This might need to be a features in our sklearn fork as it is a universal property for any supervised forest. I think we can simply leverage the

Changes proposed in this pull request:

• Add oob_samples_ as a fitted property that efficiently generates the OOB indices per tree

Before submitting

• I've read and followed all steps in the Making a pull request
  section of the CONTRIBUTING docs.
• I've updated or added any relevant docstrings following the syntax described in the
  Writing docstrings section of the CONTRIBUTING docs.
• If this PR fixes a bug, I've added a test that will fail without my fix.
• If this PR adds a new feature, I've added tests that sufficiently cover my new functionality.

After submitting

• All GitHub Actions jobs for my pull request have passed.

[codecov]

Codecov Report

Attention: 50 lines in your changes are missing coverage. Please review.

Comparison is base (4703a82) 88.84% compared to head (c1c6d64) 89.13%.

Files	Patch %	Lines
sktree/stats/forestht.py	11.53%	46 Missing ⚠️
sktree/tree/_honest_tree.py	55.55%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #200      +/-   ##
==========================================
+ Coverage   88.84%   89.13%   +0.29%     
==========================================
  Files          49       50       +1     
  Lines        4328     4492     +164     
==========================================
+ Hits         3845     4004     +159     
- Misses        483      488       +5     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[adam2392]

• Changes in Cython code were to fix the building of wheels in Windows
• Changes in GH actions workflows, examples/calibration and doc/ are used to fix the CI workflows
• All others are fair game to review
• Cython return types from intp_t to int was done in accordance with MAINT Convert int to intp_t ctype def in tree/ related code scikit-learn/scikit-learn#27546

[adam2392]

@sampan501 @PSSF23 @SUKI-O @YuxinB can you give this PR a review? Lmk if there's any questions.