[python-package] [docs] complete type annotations for scikit-learn fit() methods

[jameslamb]

Contributes to #3756.
Follow-up to #5757.

Adds type annotations on all remaining arguments to fit() methods in the scikit-learn interface.

Replaces all remaining uses of "array" or "array-like" in fit() docs with specific types, following @StrikerRUS 's suggestion from #3756 (comment).

This will hopefully make the docs a bit clearer for users, the code a bit easier to understand for contributors, and will improve mypy's ability to catch type issues (both in this project and in upstream projects that depend on lightgbm).

How I tested this

Added additional test cases in test_sklearn.py adding some other data types for init_score, sample_weight, and all the eval_* equivalents. Those tests don't cover every possible combination of all types for all arguments (that would be literally 1000s of tests), but every type in these type annotations is covered at least once by a test.

e.g., there isn't a "y is a list and init_score is a numpy array" test, but there is at least one "y is a list" test and at least one "init_score is a numpy array" test.

I also ran mypy to confirm that this PR does not add any new type-checking errors.

[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.