fix: Account for the expected kind when resolving turbofish generics

[vezenovm]

Description

Problem*

Resolves #5442

Summary*

In #5155 we make sure that we make sure we resolve against the correct type kind in various places such as when resolving trait bound generics or trait impl generics. This was not done for function calls. Now when resolving turbofish generics, rather than calling resolve_type in all cases aside type expressions, we check whether the variable we are elaborating is a function and fetch its already resolved generics and pass that in as the expected kind.

To keep the resolution logic the simple (I just zipped together the direct_generics from the function meta and the user specified turbofish generics), as zip uses the shortest iterator, I also added two checks for whether we have the correct turbofish count. We now have multiple of these checks rather than in just instantiate, but I felt it was the simplest logic and would lead to the least potential future confusion.

I also have expanded the bug identified in the issue to also include turbofish generics for method calls. I have also included this as part of my test for this PR.

Additional Context

Documentation*

Check one:

• No documentation needed.
• Documentation included in this PR.
• [For Experimental Features] Documentation to be submitted in a separate PR.

PR Checklist*

• I have tested the changes locally.
• I have formatted the changes with Prettier and/or cargo fmt on default settings.

[jfecher]

👍