DOC/MAINT: address numpydoc ignores

[lucascolley]

Across the codebase we have various numpydoc: ignore comments. PRs are always welcome to remove these ignores by adding suitable docstring content!

[crusaderky]

I disagree that the #numpydoc: ignore tags should be addressed by writing more and longer docstrings.
When they are there it's because documentation would be pointlessly redundant.

I strongly believe that numpydoc checks should be disabled for all private functions.

[lucascolley]

I think that concise, accurate docstrings can still be valuable in private code—not so much for those who wrote the code, but for others who are reading it for the first time.

I agree that adding long, redundant information is undesirable. This is low priority either way.

I strongly believe that numpydoc checks should be disabled for all private functions.

As far as I am aware, there isn't a config setting for this.