[tests][python][scikit-learn] New unit tests and maintenance

[a-wozniakowski]

This pull request:

• Includes new multioutput tests for independent subtasks and chaining.
• Includes a RandomizedSearchCV test.
• Updates dataset parameters to eliminate the FutureWarning in accordance with PEP 3102.
• Updates the GridSearchCV test such that the refit best classifier is scored on the test data-- this is also done in the RandomizedSearchCV test.

[StrikerRUS]

@a-wozniakowski Thank you very much for fixing the FutureWarning and adding more tests! Given that scikit-learn public API changes in a non-backward compatible way (#2628) quite often and passing their check_estimator test doesn't mean that we are fully compatible with the most recent API (scikit-learn/scikit-learn#15392 (comment)), I believe that having more tests for their popular tools is very important!

Please check some minor comment below:

[a-wozniakowski]

@StrikerRUS thanks for your feedback! I implemented the requested changes, and I wrote the grid/random searches such that there is an explicit train/validation/test split.

Also, scikit-learn supports **fit_params in MultiOutputClassifier and MultiOutputRegressor; see issue #15953. This seems partial progress towards issue #524, which is contained in the feature request hub #2302.

[StrikerRUS]

Thanks a lot for addressing comments! Please consider two check simplifications.

[StrikerRUS]

Thank you very much @a-wozniakowski !

[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.