chore: turn on averaging for protocol circuits metrics in CI

[TomAFrench]

Description

Problem*

Resolves

Summary*

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]

Execution Report

Program	Execution Time	%
sha256_regression	0.054s	-2%
regression_4709	0.001s	0%
ram_blowup_regression	0.575s	0%
rollup-root	0.106s	0%
rollup-block-merge	0.106s	0%
rollup-base-public	1.230s	0%
rollup-base-private	0.460s	-2%
private-kernel-tail	0.021s	0%
private-kernel-reset	0.347s	-1%
private-kernel-inner	0.077s	1%

[github-actions]

Compilation Report

Program	Compilation Time	%
sha256_regression	1.286s	-5%
regression_4709	0.787s	-2%
ram_blowup_regression	18.180s	-3%
rollup-root	3.614s	-6%
rollup-block-merge	3.736s	3%
rollup-base-public	36.320s	4%
rollup-base-private	12.380s	2%
private-kernel-tail	0.996s	-4%
private-kernel-reset	8.438s	-4%
private-kernel-inner	2.266s	8%

[github-actions]

Execution Memory Report

Program	Peak Memory
keccak256	74.65M
workspace	123.91M
regression_4709	315.98M
ram_blowup_regression	512.41M
rollup-base-public	480.66M
rollup-base-private	326.51M
private-kernel-tail	181.02M
private-kernel-reset	249.69M
private-kernel-inner	209.86M

[github-actions]

Compilation Memory Report

Program	Peak Memory
keccak256	78.11M
workspace	123.76M
regression_4709	422.96M
ram_blowup_regression	1.58G
rollup-base-public	4.11G
rollup-base-private	1.20G
private-kernel-tail	207.17M
private-kernel-reset	730.59M
private-kernel-inner	294.40M

[TomAFrench]

I've changed the external repo reports to take averages as we've reduced the compilation time significantly for much of these circuits.

[jfecher]

Now that runtimes are separated pre-SSA instead of adding another pass we should remove the resolve_is_unconstrained pass entirely and simplify away the call instruction now that runtime is stored in the dfg as well. That way we save time by needing 2 fewer passes.

Edit: #7004

[TomAFrench]

We'll also need to update codegen_if as well in that case, right? It'll quite happily create if-else structures with constant conditions atm.

[jfecher]

Right, odd I thought we checked constant conditions earlier for some reason. I'll make a fresh PR.

[TomAFrench]

@jfecher can you approve this as I've repurposed this PR to update CI (and address a failure due to noir_rsa being updated so it compiles again)