Fix type errors

[cbrnr]

Fixes #93.

[cbrnr]

In my opinion, the remaining errors/warnings (when setting "python.analysis.typeCheckingMode": "basic") are useless. These include:

• values that can be None, even if we explicitly take care of that literally in the previous line

    if classifiers_dir is None:
        classifiers_dir = get_config("classifiers_dir")
  
    target_file = Path(classifiers_dir).expanduser() / name

• things like ax[row], where Pylance says that we cannot index a Matplotlib Axes,
• and NumPy-related stuff where I cannot even read the error message because it is basically gibberish.

@hofaflo let me know if you think we should address any of the remaining errors (and if so, please show me how).

[cbrnr]

Ima turn off type checking now to stay sane.

[hofaflo]

Pyright will still complain, but I think making mypy happy is enough :)

The mkdocstrings issue is already fixed upstream (mkdocstrings/python#45), so I bumped the dependency. 5e81986 should have worked around this, but for some reason readthedocs ignored the changed path to the configuration file and kept calling mkdocs with --config-file mkdocs.yml instead of --config-file ./mkdocs.yml (see the build log) 🤷

[hofaflo]

This is ready to merge from my point of view @cbrnr :)

[cbrnr]

Looks better than expected 😄! I have a couple of questions (inline) plus the following two points:

1. Regarding the py.typed file – why is this empty file necessary (what does it do)? And if it is necessary, should it not be included as package_data?
2. I think we should add .mypy_cache to .gitignore.

[cbrnr]

@hofaflo, besides my minor comments this looks really good (I've edited my comments so that all questions are in the first comment).

[hofaflo]

Regarding the py.typed file – why is this empty file necessary (what does it do)?

PEP561 says so 🤷. I'm not totally sure but I assume it tells type checkers (such as mypy) that the package is typed.

And if it is necessary, should it not be included as package_data?

Totally, it is included in the wheels and in the source archive (I guess because its parent directory is listed in MANIFEST.in?).

I think we should add .mypy_cache to .gitignore.

For some reason my VS Code already ignored it so I didn't notice, but sure, I'll add it :D

[cbrnr]

Looks great! Ready to merge?

[hofaflo]

Yes!

[cbrnr]

Thanks @hofaflo!