[ci] [python-package] more errors from scikit-learn rearranging estimator tags

[jameslamb]

Description

scikit-learn recently merged another change to how estimator tags are handled: scikit-learn/scikit-learn#30149

As a result, the CI job here testing against scikit-learn nightlies is failing like this:

E   AttributeError: 'Tags' object has no attribute '_xfail_checks'

(build link)

Reproducible example

N/A

Environment info

Additional Comments

Relevant other discussions:

• [python-package] require scikit-learn>=0.24.2, make scikit-learn estimators compatible with scikit-learn>=1.6.0dev #6651
• [ci] [python-package] scikit-learn compatibility tests fail with scikit-learn 1.6.dev0 #6653
• [python] tests fail on scikit-learn 1.6 nightlies dmlc/xgboost#10896

[jameslamb]

The error reported above during test collection. Fixing it revealed other recent scikit-learn changes we'll have to adapt to.

As a result of scikit-learn/scikit-learn#30122, I see these failures here:

FAILED tests/python_package_test/test_sklearn.py::test_stacking_regressor - ValueError: The estimator LGBMRegressor should be a regressor
FAILED tests/python_package_test/test_sklearn.py::test_stacking_classifier - ValueError: 'final_estimator' parameter should be a classifier. Got LGBMClassifier(n_estimators=3)

(build link)