Support scikit-learn 1.9 (DTYPE removed from sklearn.tree._tree)

[MaxGhenis]

Problem

import quantile_forest raises ImportError on scikit-learn 1.9:

ImportError: cannot import name 'DTYPE' from 'sklearn.tree._tree'

scikit-learn 1.9 no longer re-exports DTYPE from sklearn.tree._tree. quantile_forest/_quantile_forest.py imports it at module load, so the whole package fails to import under scikit-learn ≥ 1.9.

Fix

DTYPE has always been np.float32, and it is used here only as the coercion dtype passed to validate_data/check_X_y (the two "dtype": DTYPE call sites). So guard the import with a fallback to np.float32:

try:
    from sklearn.tree._tree import DTYPE
except ImportError:  # scikit-learn >= 1.9 no longer re-exports DTYPE here
    DTYPE = np.float32  # the value DTYPE has always held

• Identical value — no behavior change on any scikit-learn version.
• Backward compatible — older scikit-learn still imports the real symbol.
• Forward compatible — verified import quantile_forest-equivalent against scikit-learn 1.9.0: the old import raises, the fallback yields np.float32.
• Pure-Python, single line region; numpy as np is already imported in this module.

Verification

$ python -c "import sklearn; print(sklearn.__version__)"
1.9.0
$ python -c "from sklearn.tree._tree import DTYPE"
ImportError: cannot import name 'DTYPE' from 'sklearn.tree._tree'   # before
# after this change: import succeeds, DTYPE == np.float32

Happy to add a CI entry pinning the newest scikit-learn if that's useful.

[codecov]

Codecov Report

❌ Patch coverage is 50.00000% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 98.55%. Comparing base (e193836) to head (5b19a86).

Files with missing lines	Patch %	Lines
quantile_forest/_quantile_forest.py	50.00%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #152      +/-   ##
==========================================
- Coverage   98.72%   98.55%   -0.17%     
==========================================
  Files           5        5              
  Lines        1175     1178       +3     
==========================================
+ Hits         1160     1161       +1     
- Misses         15       17       +2     

☔ View full report in Codecov by Harness.
📢 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.

[reidjohnson]

LGTM. Thank you for the contribution, @MaxGhenis!