Resolve associated types

[Veykril]

Prior we were only resolving paths until the first type was found, then discarding the result if the path wasn't fully consumed. That of course causes associated types to not resolve. Fixes #5003

[Veykril]

We got a function for this in hir_ty, I guess it would make sense to move it to hir_def?(I think it's implementation solely relies on hir_def). I have no clue in which module that would belong in that case though, unless a new utils module like it is in hir_ty
https://github.com/rust-analyzer/rust-analyzer/blob/25201b2dad7b4b0d41494e238ebf643ad7ad8cd6/crates/hir_ty/src/utils.rs#L81

[flodiebold]

I probably want to get rid of all_super_traits, but it might be nice to turn all_super_trait_refs into a query.

[flodiebold]

I'm not a fan of the duplication of logic. Actually assoc_type_shorthand_candidates is already exposed on the semantics level, so can't that be reused?

Also, I'm not sure whether this handles <T as Trait>::Type and <T>::Type?

[Veykril]

Ye the duplication isn't too nice, I'll see what I can do about that.

It doesn't handle <T as Trait>::Type yet, I already looked into that and it'll require changing a lot of ModPath params to Paths so that we don't discard the generics. Didn't feel like doing that here to keep the PR smaller.

<T>::Type isn't handled here either I believe, should be fixed if I reduce the duplication though I imagine. now handled properly

[Veykril]

Oh, I think I didn't even see assoc_type_shorthand_candiates, that should be easily reuseable.

[flodiebold]

LGTM, r=me

[Veykril]

bors r=flodiebold

[bors]

Build succeeded:

• Rust (macos-latest)
• Rust (ubuntu-latest)
• Rust (windows-latest)
• TypeScript (ubuntu-latest)
• TypeScript (windows-latest)

[Veykril]

changelog fix: Resolve associated types at the IDE layer