Remove redundant decorators on individual @overloads

[jorenham]

For some reason this is causing stubtest to crash when running it on the numpy codebase while hypothesis is installed. I'm not saying this is your fault or anything, and I'll report this issue at mypy later today, But even so, there are some other reasons for removing these decorators, so I figured I might as well clean it up.
Decorating a function has a (slight) inpact on import time, so by removing these redundant decorators, importing this module will be slightly faster.
There are also some other obvious reasons, such as readability, maintainability, seeing as it's more "DRY" this way.

[jorenham]

Oh woops, vscode seems to have messed up the formatting; sorry about that.

[Liam-DeVoe]

(./build.sh format to make it CI-compliant; we should document that in the CI job error message so it's easy for new contributors)

[jorenham]

Does this require a release note?

[Liam-DeVoe]

yeah, since we cut a release on every merged PR*. You can put something as simple as "clean up internal @overload type annotations"

*with source code changes

[Liam-DeVoe]

Thanks Joren, this looks great! I'll be curious to follow along on the mypy issue for the root cause, though it's not too surprising they assumed users wouldn't apply weird transformations on top of @overload 😄

I think CI is still a bit flaky, it's not your fault.

[Liam-DeVoe]

re: https://github.com/HypothesisWorks/hypothesis/actions/runs/18502481240/job/52723140081?pr=4564

ForwardRef.__forward_value__ was removed in 3.14: python/cpython#129465.

Not sure why it only errors on our free-threading CI job. I'll fix this soon

[Liam-DeVoe]

nvm I think fixing that will be nontrivial. I'll open an issue and get to it later. Thanks again!

[jorenham]

Thanks!