Investigate use of ``lru_cache`` on methods

[DanielNoord]

Problem

I stumbled upon this video which explains problems with using lru_cache on methods as it leaks self. I haven't watched the full video yet but watched the previous video and saw the PR at pytest here: pytest-dev/pytest#9148

This is something we're doing ourselves as well and could be a potential source of memory leaks and other problems. I think this warrants an investigation and probably a refactor.

I have also opened pylint-dev/astroid#1344 to track this.

---

I think after the investigation of internal use it might make sense to perhaps turn this into a checker? I'll report after I have finished my investigation whether that might be useful.

[asottile]

you probably also want to ban @functools.cache on methods -- glad you enjoyed my video :)

[DanielNoord]

Thanks @asottile! Wasn't sure whether if you would appreciate being tagged but I did wonder if there were other troublesome functions. I'll open an issue for cache in a couple of minutes.
I do wonder though, in the docs of functools there is an example of turning cached_property into a stacking of property and cache. Wouldn't that create the same issue?