drop pkg_resources in favour of importlib.metadata

[graingert]

re: https://setuptools.pypa.io/en/latest/pkg_resources.html

Use of pkg_resources is discouraged in favor of importlib.resources, importlib.metadata, and their backports (resources, metadata). Please consider using those libraries instead of pkg_resources.

• Closes #xxxx
• Tests added / passed
• Passes pre-commit run --all-files

[github-actions]

Unit Test Results

       12 files  ±0         12 suites  ±0   6h 4m 15s ⏱️ + 11m 58s
  2 653 tests ±0    2 571 ✔️ +1    80 💤 ±0  2 ❌  - 1 
13 021 runs  ±0  12 380 ✔️ ±0  638 💤 +1  3 ❌  - 1 

For more details on these failures, see this check.

Results for commit 266c14e. ± Comparison against base commit 7a69b5e.

♻️ This comment has been updated with latest results.

[graingert]

pandas test_register_entrypoint now looks like this: https://github.com/pandas-dev/pandas/blob/48d515958d5805f0e62e34b7424097e5575089a8/pandas/tests/plotting/test_backend.py#L47-L62

I used:

    monkeypatch.syspath_prepend(tmp_path)
    monkeypatch.setitem(sys.modules, "pandas_dummy_backend", dummy_backend)

in the pandas version of this PR because they already do a lot of mutating of sys.modules so figured a sys.path mutation wasn't so bad. Here I use a multiprocessing.Pool(1) to isolate sys.modules and sys.path changes

[crusaderky]

This looks good