Completion type match calculations for function items should consider the signature compatability

[Veykril]

No description provided.

[jmintb]

I am up for working on this :)

[Veykril]

Okay, so the bigger picture here would be to have type_match consider coercions I think (which should fix this technically). But that seems a lot more tricky to get right, so instead we should just have compute_type_match consider FnDefs and the like specially which we can do via Type::as_callable I think. Roughly, we just need to check whether both types are "callables" and then match them according to the callables instead.

[Veykril]

@jmintb I'm sorry, I just realized that was roughly what you were doing in your commit already (how embarassing) 😅 I think I misread something there yesterday when fixing up the PR. Can you just re-implement that part as a new PR (and maybe put the actual equivalence check logic behind a function in Callable (if you still want to, assuming you realized the same)?

[jmintb]

No worries, I will do as you suggest :)

[Veykril]

Fixed in #16136