Fix KeyPoints 2D boolean mask indexing (uniform-count selection)

[Copilot]

Fixes KeyPoints.__getitem__ to handle 2D boolean mask indexing where each row selects the same number of keypoints. Previously this raised IndexError; it now produces a correct (n, k, 2) result.

Use cases that now work

# Single-object confidence filter — always works (n=1, uniform trivially)
kp1[kp1.confidence > 0.5]

# Select the same subset of joints from every person in the frame
left_side = np.array([[True, False, True, False, True, ...]])  # shape (1, m)
mask = np.tile(left_side, (n, 1))   # broadcast to all n persons
kp[mask]

Limitation (documented)

A 2D mask where different rows select different numbers of True values still raises ValueError — NumPy cannot represent a jagged (n, ?, 2) array. This is the common case when using kp[kp.confidence > 0.5] with multiple objects (each person may have a different number of high-confidence keypoints). For that pattern process each object individually or zero out low-confidence entries in-place.

Changes

• KeyPoints.__getitem__ / _get_by_2d_bool_mask: New private method handles 2D np.ndarray boolean indices. Applies the mask row-by-row producing a (n, k, 2) result. Raises ValueError with a clear message if rows have different counts or mask shape mismatches.
• Shape validation: Validates mask.shape against (n_objects, n_keypoints) before processing.
• Tests: Six new parametrized cases — uniform multi-object, single-object, confidence=None, shape-mismatch ×2, all-False mask, and the canonical single-object kp[kp.confidence > 0.5] expression.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 77%. Comparing base (1291188) to head (16f0110).
⚠️ Report is 1 commits behind head on develop.

Additional details and impacted files

@@           Coverage Diff           @@
##           develop   #2188   +/-   ##
=======================================
  Coverage       77%     77%           
=======================================
  Files           62      62           
  Lines         7615    7637   +22     
=======================================
+ Hits          5881    5903   +22     
  Misses        1734    1734           

🚀 New features to boost your workflow:

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

[Copilot]

Pull request overview

Fixes KeyPoints.__getitem__ behavior when indexing with a 2D boolean mask (e.g., keypoints[keypoints.confidence > 0.5]) by adding explicit handling for (n, m) boolean masks and adding tests for the new behavior.

Changes:

• Added a dedicated __getitem__ branch to support 2D boolean mask filtering of keypoints per object, with validation for shape consistency.
• Added parametrized tests covering valid 2D-mask filtering and expected ValueError cases (row/column mismatch and unequal per-row selection counts).

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
src/supervision/key_points/core.py	Implements 2D boolean mask handling in KeyPoints.__getitem__.
tests/key_points/test_core.py	Adds test cases for 2D boolean mask indexing behavior and error conditions.