[red-knot] Support assert_type

[InSyncWithFoo]

Summary

See #15103.

Test Plan

Markdown tests and unit tests.

[github-actions]

ruff-ecosystem results

Linter (stable)

✅ ecosystem check detected no linter changes.

Linter (preview)

✅ ecosystem check detected no linter changes.

[InSyncWithFoo]

This is largely a proof-of-concept. is_equals_to() is not a thing in the specification, and I'm quite sure its name is not the best.

[sharkdp]

Thank you very much for working on this. I added a few initial review comments.

[sharkdp]

Can you go into a bit more detail in how this is different from is_equivalent_to? The main difference seems to be that this can also handle non-fully-static types and return true for something like Any ~ Any? If so, could it handle those gradual types (+ intersections/unions/tuples) and then fall back to is_equivalent_to for fully static types?

[AlexWaygood]

I think it's quite possible that a function like this could help fix #14899, FWIW, so it doesn't seem like an unreasonable addition to me. I'm also not sure what name would clearly distinguish it from Type::is_equivalent_to(), though 😆

[sharkdp]

To be clear, I was not suggesting that this function is not useful. I just think there's some overlap with is_equivalent_to, and I'd like to understand it and possibly make use of it.

[InSyncWithFoo]

Another notable exception is type[Any] != type != type[object]. I'm not too sure about the former, but I think it is better to be strict.

Also, as it currently is, int | Unknown | Any != int | Any. This is supposedly due to Unknown not being foldable into Any, but int | Any is just Any; could such union types even be emitted in the first place?