fix(ssa): Use number of SSA instructions for the Brillig unrolling bytecode size limit

[vezenovm]

Description

Problem*

Pulls a change out of #7236 that was causing panics with less aggressive inliner settings. Comment for reference: #7236 (comment).

I labelled this PR a fix rather than a chore as it does prevent possible panics.

Summary*

After performing an unroll in the Brillig runtime we have a setting that lets us set an upper bound on the bytecode size increase percentage. However, we have certain intrinsics (e.g. DerivePedersenGenerators) that we expect to entirely compile-time and fully resolved by Brillig gen.

Another possible fix would be to set any method which calls derive_generators (such as pedersen_hash_with_separator) as #[inline_always]. However, we still do not have a guarantee that the separator has been resolved to a constant. Let's say we know derive_generators is called by pedersen_hash_with_separator. Even if pedersen_hash_with_separator is always inlined into its parent we cannot guarantee that the input for the separator comes from the parent's function parameter. This issue will exist with any other intrinsics we expect to be fully resolved by compile-time.

By using the number of SSA instructions we can also avoid the extra compilation to Brillig as well as having to expose extra Brillig gen logic to the SSA passes. This gives us a better separation of concerns.

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.

[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: f6bf769	Previous: 15bbaa8	Ratio
noir-lang_schnorr_	1 s	0 s	+∞
noir-lang_noir_check_shuffle_	1 s	0 s	+∞
noir-lang_mimc_	1 s	0 s	+∞

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

CC: @TomAFrench

[jfecher]

This seems much simpler than going off of brillig instructions