[python-package] fix mypy errors in Dataset construction

[jameslamb]

Contributes to #3756.
Contributes to #3867.

Fixes the following errors from mypy

basic.py:1922: error: Argument 1 to "__init_from_list_np2d" of "Dataset" has incompatible type "list[Any] | list[Sequence] | list[ndarray[Any, Any]]"; expected "list[ndarray[Any, Any]]"  [arg-type]
basic.py:1924: error: Argument 1 to "__init_from_seqs" of "Dataset" has incompatible type "list[Any] | list[Sequence] | list[ndarray[Any, Any]]"; expected "list[Sequence]"  [arg-type]
basic.py:2874: error: Argument 1 to "_yield_row_from_seqlist" of "Dataset" has incompatible type "list[Any] | list[Sequence] | list[ndarray[Any, Any]]"; expected "list[Sequence]"  [arg-type]

These errors all come from the fact that you can't use an expression like

if isinstance(l, list) and all(isinstance(x, np.ndarray) for x in l):
    # ...

To help mypy understand which type from a union like Union[List[np.ndarray], List[Sequence]] a particular code block is working with.

As described in https://mypy.readthedocs.io/en/stable/type_narrowing.html#typeguards-with-parameters, in Python 3.10 mypy introduced a feature to do exactly that. If you set up a function like f(l) -> typing.TypeGuard[List[np.ndarray]], that tells the type checker:

• f() returns a bool
• if f(l) returns True, then l must be a List[np.ndarray]

This PR proposes adding such guards to lightgbm. This will help type checkers (like mypy in this project's CI and others in users' IDEs) to catch errors deeper in the code paths involving lists of numpy arrays and lists of Sequence objects.

[jmoralez]

Nice!

[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.