MGCA: Avoid code duplication between HIR ty lowering and elsewhere

[BoxyUwU]

Currently the HIR ty lowering logic for direct const arguments has a lot of overlap with code elsewhere in the compiler. For example:

• We have logic in HIR type check for resolving type relative paths, and also logic in HIR ty lowering
• We have logic in HIR type check for type checking struct expressions, and we have very similar logic in HIR ty lowering for validating that a direct struct expression has the right amount of fields.

I'm not sure what the right way to avoid this code duplication is, but we should fix this before this code duplication becomes something that affects stable code.

[0xllx0]

@rustbot claim