fix: hoist and then deduplicate

[guipublic]

Description

Problem*

Resolves #10030

Summary*

Following the great investigations from @aakoshh in the issue's comments, one can see that this is a reference count issue.
When there is a make_array in block A, and another identical one in block B;
if block A dominates B, we can safely remove the second instruction. In that case, the reference count is increased, in case the redundant array was mutated afterwards.
However, when there is no dominator relation between A and B, the second instruction is removed and inserted in a common dominator, without increasing the reference count.
We then rely on a next constant folding to remove the instruction from A.

The fix consists of not removing the second instruction, but still inserting a copy of the instruction in the common dominator. The next constant folding will properly remove instructions in A and B and insert reference counts as needed.

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 'Execution Memory'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.20.

Benchmark suite	Current: 6f72c0a	Previous: cb5c0ed	Ratio
rollup-checkpoint-root-single-block	9680 MB	1030 MB	9.40
rollup-checkpoint-root	9680 MB	1030 MB	9.40
rollup-tx-base-private	1360 MB	450.76 MB	3.02
rollup-tx-base-public	6950 MB	465.63 MB	14.93

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

CC: @TomAFrench

[github-actions]

Changes to Brillig bytecode sizes

Generated at commit: e023c53d226e2070e2f0d8f05bea01aaff02c975, compared to commit: aaa27ba1adda8ddefeea0dbdc1660ec7d0deb4cb

🧾 Summary (10% most significant diffs)

Program	Brillig opcodes (+/-)	%
lambda_from_array_inliner_max	+120 ❌	+3.09%
lambda_from_array_inliner_min	+120 ❌	+3.09%
lambda_from_array_inliner_zero	+120 ❌	+3.09%
lambda_from_dynamic_if_inliner_zero	+12 ❌	+2.42%

---

Full diff report 👇

Program	Brillig opcodes (+/-)	%
lambda_from_array_inliner_max	4,008 (+120)	+3.09%
lambda_from_array_inliner_min	4,008 (+120)	+3.09%
lambda_from_array_inliner_zero	4,008 (+120)	+3.09%
lambda_from_dynamic_if_inliner_zero	508 (+12)	+2.42%
lambda_from_dynamic_if_inliner_min	516 (+12)	+2.38%
lambda_from_dynamic_if_inliner_max	620 (+12)	+1.97%
regression_9496_inliner_max	540 (+6)	+1.12%
regression_9496_inliner_min	540 (+6)	+1.12%
regression_9496_inliner_zero	540 (+6)	+1.12%
reference_counts_inliner_max_inliner_max	744 (+6)	+0.81%
reference_counts_inliner_min_inliner_max	744 (+6)	+0.81%
reference_counts_inliner_0_inliner_max	744 (+6)	+0.81%
reference_counts_slices_inliner_0_inliner_max	773 (+6)	+0.78%
nested_array_dynamic_inliner_min	1,333 (+6)	+0.45%
nested_array_dynamic_inliner_zero	1,333 (+6)	+0.45%
nested_array_dynamic_inliner_max	1,607 (+6)	+0.37%
nested_array_in_slice_inliner_min	809 (+3)	+0.37%
nested_array_in_slice_inliner_zero	809 (+3)	+0.37%
slices_inliner_max	1,685 (+6)	+0.36%
reference_counts_inliner_max_inliner_zero	881 (+3)	+0.34%
reference_counts_inliner_0_inliner_zero	881 (+3)	+0.34%
reference_counts_inliner_min_inliner_zero	881 (+3)	+0.34%
nested_array_in_slice_inliner_max	1,001 (+3)	+0.30%
reference_counts_slices_inliner_0_inliner_zero	1,002 (+3)	+0.30%
slices_inliner_zero	1,645 (+3)	+0.18%
slices_inliner_min	2,180 (+3)	+0.14%
uhashmap_inliner_max	11,263 (+3)	+0.03%
hashmap_inliner_max	18,577 (+3)	+0.02%

[github-actions]

Changes to number of Brillig opcodes executed

Generated at commit: e023c53d226e2070e2f0d8f05bea01aaff02c975, compared to commit: aaa27ba1adda8ddefeea0dbdc1660ec7d0deb4cb

🧾 Summary (10% most significant diffs)

Program	Brillig opcodes (+/-)	%
regression_9496_inliner_max	+6 ❌	+1.18%
regression_9496_inliner_min	+6 ❌	+1.18%
regression_9496_inliner_zero	+6 ❌	+1.18%

---

Full diff report 👇

Program	Brillig opcodes (+/-)	%
regression_9496_inliner_max	514 (+6)	+1.18%
regression_9496_inliner_min	514 (+6)	+1.18%
regression_9496_inliner_zero	514 (+6)	+1.18%
nested_array_in_slice_inliner_max	1,199 (+3)	+0.25%
lambda_from_array_inliner_max	1,242 (+3)	+0.24%
lambda_from_array_inliner_min	1,242 (+3)	+0.24%
lambda_from_array_inliner_zero	1,242 (+3)	+0.24%
nested_array_dynamic_inliner_max	2,529 (+6)	+0.24%
nested_array_in_slice_inliner_min	1,298 (+3)	+0.23%
nested_array_in_slice_inliner_zero	1,298 (+3)	+0.23%
nested_array_dynamic_inliner_min	2,731 (+6)	+0.22%
nested_array_dynamic_inliner_zero	2,731 (+6)	+0.22%
slices_inliner_max	2,532 (+3)	+0.12%
hashmap_inliner_max	57,057 (+46)	+0.08%
uhashmap_inliner_max	143,336 (+46)	+0.03%

[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: 6df8558	Previous: aaa27ba	Ratio
rollup-checkpoint-merge	0.004 s	0.003 s	1.33

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

CC: @TomAFrench

[aakoshh]

🙌

[aakoshh]

I guess the regressions make sense, hopefully #10019 can reclaim some of it.
cc @vezenovm

[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: 8ff2417	Previous: 46df8f0	Ratio
test_report_AztecProtocol_aztec-packages_noir-projects_noir-protocol-circuits_crates_blob	374 s	276 s	1.36
test_report_zkpassport_noir-ecdsa_	2 s	1 s	2
test_report_zkpassport_noir_rsa_	2 s	1 s	2

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

CC: @TomAFrench