Type inference 2.0 - #21795

By lack of a better title, I named this PR 'Type inference 2.0', because it makes a lot of changes (improvements) to our implementation of type inference for Rust. At a high level, a lot of code is moved from the Rust codebase to the shared type inference library (in preparation for unified/Swift), and there is now a very clear distinction between bottom-up type inference and top-down (contextual) type inference.

Before this PR

• The shared type inference library contained the core functionality for propagating type information through for example function calls, but all logic for mapping AST nodes to types was done outside of the library.
• Rust type inference allowed for types to propagate bidirectionally (as a classical constrained-based implementation would), but since this could easily lead to explosions in inferred types, advanced logic existed for inferring types with certainty, and this logic also tried to infer certain types for calls.

Note for the reviewer

As usual, commit-by-commit reviewing is encouraged. The first commit (which compiles and works) moves a bunch of logic around in the Rust implementation, which is then removed in the subsequent commit (which doesn't compile). I found that doing it like this resulted in a cleaner diff on the last commit, and it also makes it more clear which parts of the logic are now handled by shared code.

Impact

Future work

• Handle more AST constructs in the shared library, such as patterns (should be relatively straightforward).
• Shared logic for type-based overload resolution; logic currently exists for Rust, and perhaps some of this logic can be shared.