ENH Cleanup multiview API and enable oblique multiview

[adam2392]

Changes proposed in this pull request:

• Supersedes: ENH Cleanup multiview API and enable oblique multiview #247
• Implements multiview oblique random forest
• Removes kwarg apply_max_features_per_feature_set in the multiview API. This is done in favor of the user explicitly specifying the max_features array

Summary

The multiview oblique splitter differs from oblique splits and axis-aligned split implementation in a few ways:

1. It allows a different max_features argument per feature set.
2. It definitely samples a non-trivial projection vector for each mtry. So if mtry for feature set A is 5, then it will have 5 possible projection vectors
3. It allows user to turn on/off projection vector sampling that goes across feature sets. For example, if try for feature set A is 5, with cross-feature-set-sampling, then it will sample 5 projection vectors. Each projection vector will include a feature from feature set A, but also possibly a feature from other feature sets B, C, …etc.

The algorithmic differences are visualized here: https://output.circle-artifacts.com/output/job/9e34e7cb-37d1-488c-bd7b-beba5689401b/artifacts/0/dev/auto_examples/splitters/plot_multiview_axis_aligned_splitter.html#sphx-glr-auto-examples-splitters-plot-multiview-axis-aligned-splitter-py

Reviewing Notes

The main changes to review are: sktree/tree/_oblique_splitter.pyx and specifically the sample_proj_mat function within MultiViewObliqueSplitter.

TODO:

1. Add unit-tests for highly-correlated features that should improve over MultiViewDecisionTreeClassifier.

[codecov]

Codecov Report

❌ Patch coverage is 79.48718% with 32 lines in your changes missing coverage. Please review.
✅ Project coverage is 78.89%. Comparing base (e8c7de5) to head (90320b1).
⚠️ Report is 40 commits behind head on main.

Files with missing lines	Patch %	Lines
treeple/tree/_multiview.py	76.81%	18 Missing and 14 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #265      +/-   ##
==========================================
+ Coverage   78.66%   78.89%   +0.22%     
==========================================
  Files          24       24              
  Lines        2264     2369     +105     
  Branches      417      434      +17     
==========================================
+ Hits         1781     1869      +88     
- Misses        352      362      +10     
- Partials      131      138       +7     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.