[python-package] simplify scikit-learn 1.6+ tags support

[jameslamb]

Follow-up to #6718

In that PR, I'd skipped the method resolution order ("MRO") for __sklearn_tags__() in the scikit-learn estimators, to provide compatibility with earlier versions of that library. During development, we switched to having e.g. LGBMClassifier.__sklearn_tags__() raise an AttributeError if run on a too-new version of scikit-learn, making that unnecessary... but never went back to just using inheritance to get __sklearn_tags__().

This fixes that.

Notes for Reviewers

Inspired by dmlc/xgboost#11021 (comment)

[jameslamb]

This is necessary even though it doesn't do anything, because of this check in scikit-learn:

E TypeError: Estimator LGBMRegressor has defined either _more_tags or _get_tags, but not __sklearn_tags__. If you're customizing tags, and need to support multiple scikit-learn versions, you can implement both __sklearn_tags__ and _more_tags or _get_tags. This change was introduced in scikit-learn=1.6

https://github.com/scikit-learn/scikit-learn/blob/fba028b07ed2b4e52dd3719dad0d990837bde28c/sklearn/utils/estimator_checks.py#L4459-L4465

Added in scikit-learn/scikit-learn#30268

[StrikerRUS]

LGTM, except one comment below. Thank you for coming back and simplifying this!

[github-actions]

This pull request has been automatically locked since there has not been any recent activity since it was closed.
To start a new related discussion, open a new issue at https://github.com/microsoft/LightGBM/issues including a reference to this.