chore(ACIRgen): always compute array offset

[asterite]

Description

Problem

No issue.

Summary

Revival of #10099 because I still think that PR is correct. I'll try it in Aztec-Packages again too. Maybe it wasn't working well because of other bugs we had back then, not sure.

Additional Context

User Documentation

Check one:

• No user documentation needed.
• Changes in docs/ included in this PR.
• [For Experimental Features] Changes in docs/ 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]

Changes to circuit sizes

Generated at commit: 1f8e27ae038f151fb6e76f06d7812a3f8ea9b6aa, compared to commit: 939f7c7e727fba20b7daf58856a6e3e6d4e34960

🧾 Summary (10% most significant diffs)

Program	ACIR opcodes (+/-)	%	Circuit size (+/-)	%
regression_struct_array_conditional	0 ➖	0.00%	-1 ✅	-0.03%
regression_11048	-2 ✅	-5.26%	-2 ✅	-0.07%

---

Full diff report 👇

Program	ACIR opcodes (+/-)	%	Circuit size (+/-)	%
hashmap	29,356 (-1)	-0.00%	90,309 (-1)	-0.00%
regression_struct_array_conditional	68 (0)	0.00%	3,215 (-1)	-0.03%
regression_11048	36 (-2)	-5.26%	2,886 (-2)	-0.07%

[github-actions]

⚠️ Performance Alert ⚠️

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

Benchmark suite	Current: 3e5acf7	Previous: 939f7c7	Ratio
rollup-checkpoint-merge	0.003 s	0.002 s	1.50

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

CC: @TomAFrench

[jfecher]

#10085 mentions we need to add test coverage due to the regression in aztec-packages. Did we ever add this? CI is green here but was before as well.

[asterite]

Ah, no, because I think I never figured out what was the issue, or maybe I didn't investigate what the issue was, I can't remember.

[asterite]

Well, it still fails: AztecProtocol/aztec-packages#21061

I still don't understand why, though, but for now I'll close this. I might also try to reproduce the issue locally.