[ruff] Teach autofix for used-dummy-variable about TypeVars etc. (RUF052)

[AlexWaygood]

Summary

For some bindings, such as assignments where the right-hand-side is a TypeVar constructor call, renaming the binding can result in invalid code. For example, this diff that renamer.rs might make would take code that mypy (or any other spec-compliant type checker) would have accepted, and turns it into something a type checker will reject:

  from typing import TypeVar

- _T = TypeVar("_T")
+ T = TypeVar("_T")

This is because the first argument to the TypeVar constructor must be the same as the variable the typevar is being bound to. Detecting this issue in general is impossible (there are lots of classes where this pattern is required), but it seems reasonable to special-case several standard-library classes where this pattern is required.

Fixes #14798

Test Plan

Added some more fixtures to RUF052 to check that the new logic in renamer.rs works as expected.

[github-actions]

ruff-ecosystem results

Linter (stable)

✅ ecosystem check detected no linter changes.

Linter (preview)

✅ ecosystem check detected no linter changes.

[MichaReiser]

Do you know what editors do when you rename a variable that's a TypeVar declaration?

[MichaReiser]

This is very cool! Nice work

[AlexWaygood]

Do you know what editors do when you rename a variable that's a TypeVar declaration?

VSCode just does this, so we'll do a better job than them with this change ;)

-_T = TypeVar("_T")
+T = TypeVar("_T")