fix: error on unused generic in trait impl

[asterite]

Description

Problem

Resolves #8394

Summary

In addition to applying the same logic that existed for impls to trait impls, some code stopped compiling when it was generated via macros. These end up using resolved types which weren't handled before, and now they are.

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.

[asterite]

Asking for a review because I'm not sure how to fix HashMap and UHashMap in the standard library so this error doesn't trigger... but maybe one of you know!

[jfecher]

Asking for a review because I'm not sure how to fix HashMap and UHashMap in the standard library so this error doesn't trigger... but maybe one of you know!

Rust's BuildHasher uses an associated type to avoid this: https://doc.rust-lang.org/std/hash/trait.BuildHasher.html. So we should do the same since we don't want to allow multiple impls of BuildHasher for the same input type either.

[asterite]

Cool, thank you! I'll try to do that.

[asterite]

Hmm, I think it still won't work. In Rust it's something like this:

pub trait BuildHasher {
    type H: Hasher;
}

so we know the type H is a Hasher. Right now in Noir that doesn't work but we can do this:

pub trait BuildHasher {
    type H;

    fn build_hasher(self) -> H
    where
        H: Hasher;
}

but it's not exactly the same. Whenever we have something of type B we have to constrain that to B: BuildHasher<H>, H: Hasher over and over again, but sometimes we can't introduce that type H, for example in a trait implementation where we can't add new where clauses.

Is this something that should be implemented for Noir 1.0? Otherwise once we do this it's going to be a large breaking change.

[github-actions]

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'Test Suite Duration'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.20.

Benchmark suite	Current: ef04db6	Previous: 79fb482	Ratio
test_report_AztecProtocol_aztec-packages_noir-projects_noir-protocol-circuits_crates_private-kernel-lib	2 s	1 s	2

This comment was automatically generated by workflow using github-action-benchmark.

CC: @TomAFrench

[TomAFrench]

I updated the bigcurve dep on bignum

[asterite]

Ah! Mystery solved 😄 Thanks!

[asterite]

This is ready for review now.