[ruff] Implement post-init-default (RUF033)

[tjkuson]

Summary

Implements the post-init-default (RUF033) rule, which reports a diagnostic upon any __post_init__ method with argument default. Has a sometimes available unsafe autofix. For example,

 from dataclasses import InitVar, dataclass


 @dataclass
 class Foo:
     """A very helpful docstring."""

+    baz: InitVar[str] = "something"
     bar: InitVar[int] = 1

-    def __post_init__(self, bar: int, baz: str = "something") -> None: ...
+    def __post_init__(self, bar: int, baz: str) -> None: ...

Closes #13128

Test Plan

cargo nextest run

[github-actions]

ruff-ecosystem results

Linter (stable)

✅ ecosystem check detected no linter changes.

Linter (preview)

✅ ecosystem check detected no linter changes.

[AlexWaygood]

Sorry for reviewing a draft PR... couldn't resist! Looks great.

[codspeed-hq]

CodSpeed Performance Report

Merging #13192 will improve performances by 5.89%

_{Comparing tjkuson:post-init-defaults (7b21ade) with main (0f85769)}

Summary

⚡ 1 improvements
✅ 31 untouched benchmarks

Benchmarks breakdown

	Benchmark	main	tjkuson:post-init-defaults	Change
⚡	linter/all-with-preview-rules[numpy/globals.py]	866.2 µs	818 µs	+5.89%

[AlexWaygood]

Thanks! I pushed a couple of simplifications, but this was overall an excellent PR

[AlexWaygood]

I made this run on StmtFunctionDef nodes rather than StmtClassDef nodes, as this allows us to access checker.semantic().current_scope() to see what other symbols have been bound in the class. This enables us to see much more simply and more accurately whether it's safe to add a new InitVar field to the class body in the fix

[AlexWaygood]

Rather than iterate through all statements in the class body until we get the first non-docstring statement, I changed it so that we just insert the InitVar symbol in the line before the __post_init__ definition. It seems much simpler and less error-prone as we already know that we're past the docstring in this case