Typecheck typeshed's code with pyright

[Avasam]

When it comes to typeshed's own code, it is tested with mypy but not with pyright. A bit odd given how strictly the stubs are tested against pyright. Making typeshed, imo, a bit of a "poorly shod shoemaker" on that aspect.
Pyright specifically complements mypy's missing features well.

But this is not just about typeshed itself. As it is a pillar in Python's typing ecosystem, problems caught by it, and improvements made thanks to it (especially those pushing the limitations of the current type system) have bigger repercussions on the ecosystem as a whole.

Tangible improvements:
All of the PRs and issues mentioned here are directly related, and opened due, to me trying to run pyright on typeshed's own code and working on this PR:

• Some new type issues or can be caught sooner (see comment https://github.com/python/typeshed/pull/9694/files#r1109192407)
• More chance of a pyright bug or limitation being reported by a typeshed maintainer
  • Error code for subclassing final microsoft/pyright#4483
  • Feature Request: support # pyright: ignore not at the start of a comment microsoft/pyright#4259
  • How to report unnecessary pyright: ignore comments, but not type: ignore ? microsoft/pyright#4243
  • Empty collection's subtype evaluates to Unknown when in tuple passed to Iterable with TypeVar parameter microsoft/pyright#4244
  • typing_extension.Self infers the wrong type from __new__ microsoft/pyright#4668
• Improve typeshed's own type-safety
  • See .py changes in this PR
  • The need and use for a centralized matadata file parser is something I had mentioned not too long before AlexWaygood started on it Resolve stubtest allowlist entries for ldap3 #9532 (comment)
  • Deduplicate pyright version #9299
  • Simplify pyrightconfig by removing redundant keys #9293
  • Simplify pyrightconfig.json , add missing comments and extra strict settings in all pyright configs #9714
  • Check for unused pyright: ignore and differentiate from mypy ignores #9397
  • Should the dict.get overload with default value not be generic? #9155
  • Bump various test dependencies #9549
  • A more accurate termcolor.colored fallback #9435
  • Fix potentially unbound issues in stubsabot #9754
• Spot missing and problematic types in libraries used by typeshed. Which are mostly quite common libraries that could really serve being fully typed
  • Update pkg_resources-stubs for use in pytype_test #9747
  • Type Loader parameter in PyYAML/yaml #9752
  • Expand pathspec range hadialqattan/pycln#186
  • Add py.typed marker cpburnz/python-pathspec#67
  • Fix unknown 'Self' cpburnz/python-pathspec#71
  • Fix partially unknown PathLike type cpburnz/python-pathspec#75
  • pytype is not py.typed google/pytype#1325
  • Breaking # pyright: ignore[] used with # type: ignore[] comments on short lines psf/black#3399
  • Missing type argument on IO python-poetry/tomlkit#272
  • Possible issue with PyYAML's load() #2886 (comment)
• And possibly more

(I waited until most of these got resolved before bringing this PR)

[github-actions]

According to mypy_primer, this change has no effect on the checked open source code. 🤖🎉

[AlexWaygood]

I agree with everything Jelle said; here's a few more pieces of feedback

[Avasam]

Thanks for the reviews and patience! This started absolutely huge that I split off in its own chunks over time. And there was some artefacts of that leftover.
I self-reviewed initially, but I guess I should've done it a few more times along the way. 😅