[red-knot] Infer generator expressions

[dylwil3]

Adds a Generator type and infers the type of a generator expression as Generator[T, None, None] where T is the inferred type of the first comprehension.

Related to #12701.

[dylwil3]

This is my first one of these, so apologies if I'm way off!

[github-actions]

ruff-ecosystem results

Linter (stable)

✅ ecosystem check detected no linter changes.

Linter (preview)

✅ ecosystem check detected no linter changes.

[carljm]

Hey, thanks so much for the PR! So I should have added more color to #12701 to clarify some of these better; this is an example of one where we're kinda blocked on adding support for generics in order to do this correctly. The generator type should be derived from its definition in typeshed rather than added as a special case in the Type enum. And understanding its definition in typeshed requires that we understand type variables and generic classes, which we don't have. Sorry this wasn't clear in the issue, I've updated the text of the issue now to mark the ones blocked on this!

Going to close this since I think pretty much everything here would look different with the typeshed/generics approach, sorry about that! If you're interested in tackling an open expression kind, there are gaps in our unary/binary expression support, or for something a bit more complex, lambda expressions might be interesting. @chriskrycho is currently working on call expression inference, which will unlock a bunch more expression kinds where we should be resolving them as dunder method calls. Feel free to ping us on Discord or in the issue if you want to get a bit of direction before starting work on an issue!

[dylwil3]

Not a problem! Sorry for spamming the PRs 😅 I'll coordinate on the issue/Discord before taking another stab. Thanks!