chore: move delayed_type_checks field from NodeInterner to TypeChecker

[ghost]

Problem

The current implementation of the node interner in the codebase appears to be handling multiple responsibilities and exhibiting signs of complexity.

PR Checklist

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

[jfecher]

Looks good! I'm not sure why we didn't do it this way to begin with.

[ghost]

Hey, @jfecher! How do you think there is a sense to move fields into separate structures:

• next_type_variable_id
• globals
• primitive_methods, struct_methods

[jfecher]

In general anything that is used across both name resolution and type inference (+ potentially monomorphization) belongs in the node interner. Since the interner is the general context object that accompanies the HIR.

• next_type_variable_id - also referenced by name resolution when translating types, so it is fine in the node interner.
• globals - I believe these are referenced in name resolution until they're later inlined during monomorphization. So they should be fine as well.
• primitive_methods/struct_methods - they're populated during name resolution and resolved during type checking so the node interner is fine for these as well.

So in summary, I think each of these are fine where they are.