Fix str subclass validation for enums

[sydney-runkle]

Fix pydantic/pydantic#9265
Fix pydantic/pydantic#9375

This is kind of gross, but preserves behavior from 2.6 that I think is valid.

[codspeed-hq]

CodSpeed Performance Report

Merging #1273 will not alter performance

_{Comparing enum_fixes (cad59ca) with main (b777774)}

Summary

✅ 155 untouched benchmarks

[davidhewitt]

Do we have the same problem for int subclasses? (I think bools cannot be subclassed.)

I wonder if there's a more general structural issue here (like e.g. maybe lookup.validate should take a strict argument)?

[sydney-runkle]

Alright, so we want to move forward with this by modifying how we validate enums (moving away from the literal validation logic). Definitely something we still want to fix, but going to wait on 2.7.2 for this one 👍

[samuelcolvin]

otherwise I don't love it, but I prefer it to copy & pasting lots of literal logic.

[davidhewitt]

Looks good to me!