DOC: Fixed PR01 for pandas.Series.cat

[03darius]

Ref #58504

Fixes PR01 for

-i "pandas.Series.cat.add_categories PR01,PR02" \
-i "pandas.Series.cat.as_ordered PR01" \
-i "pandas.Series.cat.as_unordered PR01" \
-i "pandas.Series.cat.remove_categories PR01,PR02" \
-i "pandas.Series.cat.remove_unused_categories PR01" \
-i "pandas.Series.cat.rename_categories PR01,PR02" \
-i "pandas.Series.cat.reorder_categories PR01,PR02" \
-i "pandas.Series.cat.set_categories PR01,PR02" \

I couldn't figure out how to fix PR02. If I'm getting it right, numpydoc detects PR02 errors if there are parameters described in the docstring that are not in the method signature. I dug into how the docstring validation works and found that it uses the signature of f defined in pandas/core/accessor.py instead of the actual signature of the methods defined in pandas/core/arrays/categorical.py

For reference, here's that delegator:

def _create_delegator_method(name: str):
    def f(self, *args, **kwargs):
        return self._delegate_method(name, *args, **kwargs)

    f.__name__ = name
    f.__doc__ = getattr(delegate, accessor_mapping(name)).__doc__

    return f

So PR02 is always detected if the method has any parameters. How should this be handled?

[Aloqeely]

You're right, this seems to be a problem on pandas' end. If you can investigate and open a separate PR to fix the problem that would be really appreciated!