Use ruff for linting and formatting

[jarrodmillman]

A couple of issues:

1. We don't appear to be testing BadExamples in numpydoc/tests/test_validate.py.
2. I had to fix the line numbers (see 2e6d88b) to get the prerelease tests to pass, but it wasn't necessary for the standard tests. For the standard tests we run pytest -v --pyargs numpydoc. For the prerelease test we run pytest -v --pyargs ..

numpydoc/tests/hooks is in the source directory:

$ ls -1 numpydoc/tests/
hooks
test_docscrape.py
test_full.py
test_main.py
test_numpydoc.py
test_validate.py
test_xref.py
tinybuild

but isn't in the installed package

$ ls -1 ~/.venv/nx/lib64/python3.11/site-packages/numpydoc/tests/
test_docscrape.py
test_full.py
test_main.py
test_numpydoc.py
test_validate.py
test_xref.py
tinybuild

I ignore a large number of linting checks because they don't have automated fixes (or their automated fixes are unsafe). I think we should eventually enable them one by one and implement the fixes in follow-up PRs. I would like to keep them separate because they will require more attention and careful review than these changes. I would like to get this PR in relatively quickly because I think moving to ruff is worth it. It is much faster, implements multiple tools, is being rapidly developed, and has been adding features specific for our community.

[stefanv]

I did not know a function with only a docstring is valid!

[stefanv]

Looks good to me; I agree that it's better to get the toolchain in place quickly, and to the iteratively refine.