DOC: Validation of docsting validates Python type doc

[datapythonista]

Our script scripts/validate_docstrings.py validates that docstrings in our code follow certain standards (the right parameters are documented, the docstring contain examples, capitalization and punctuation...).

There is currently one case when the validation is incorrect. See this (hypothetical) example:

class Series:
    ndim = 1

If we execute ./scripts/validate_docstrings.py pandas.Series.ndim, we'd expect the script to detect that ndim doesn't have a docstring. But instead, the script is validating the docstring of the type int (the doc of the scalar 1).

This is happening in practice for offset, and is preventing us to validate some error types to the CI. See: #26982 (comment)

We should detect this case in the script, and not validate the docstrings of attributes.

[davefol]

I'm willing to take this on.

[datapythonista]

@galuhsahid looks like we've got an issue for the problem we discussed.