feat: Don't increment parameter ref counts at all

[jfecher]

Description

Problem*

Summary*

We may not need to increment reference counts in parameters at all since #6568 eliminates them for arrays, and previously we had eliminated them for references as well, though this was reverted. This PR is speculation that the reversion was only needed because we forgot to also remove the dec_rc in that case, leading to reference counts drifting down too far.

Additional Context

This is passing locally but I think we're missing a inc_rc when dereferencing that is now required with this change. Adding it may be more expensive than just keeping inc_rc for reference parameters though.

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]

Changes to number of Brillig opcodes executed

Generated at commit: b2b1c32d8184712b920d69362263a1619dd9c4d4, compared to commit: b0245811bfd84e0bf3559aa1e2f37ec52d08691e

🧾 Summary (10% most significant diffs)

Program	Brillig opcodes (+/-)	%
array_sort	-42 ✅	-7.49%
brillig_rc_regression_6123	-39 ✅	-12.75%

---

Full diff report 👇

Program	Brillig opcodes (+/-)	%
fold_2_to_17	1,090,791 (+21,730)	+2.03%
bench_2_to_17	588,206 (+11,425)	+1.98%
fold_numeric_generic_poseidon	5,103 (+52)	+1.03%
no_predicates_numeric_generic_poseidon	5,103 (+52)	+1.03%
poseidon2	695 (+1)	+0.14%
regression_5252	908,660 (+72)	+0.01%
reference_counts	306 (-3)	-0.97%
uhashmap	144,588 (-1,752)	-1.20%
hashmap	51,261 (-1,675)	-3.16%
array_sort	519 (-42)	-7.49%
brillig_rc_regression_6123	267 (-39)	-12.75%

[github-actions]

Changes to Brillig bytecode sizes

Generated at commit: b2b1c32d8184712b920d69362263a1619dd9c4d4, compared to commit: b0245811bfd84e0bf3559aa1e2f37ec52d08691e

🧾 Summary (10% most significant diffs)

Program	Brillig opcodes (+/-)	%
hashmap	-753 ✅	-3.80%
brillig_rc_regression_6123	-27 ✅	-14.75%

---

Full diff report 👇

Program	Brillig opcodes (+/-)	%
regression_5252	4,519 (-6)	-0.13%
reference_counts	337 (-3)	-0.88%
array_sort	289 (-3)	-1.03%
no_predicates_numeric_generic_poseidon	737 (-11)	-1.47%
fold_numeric_generic_poseidon	737 (-11)	-1.47%
poseidon2	334 (-6)	-1.76%
bench_2_to_17	327 (-6)	-1.80%
uhashmap	12,922 (-264)	-2.00%
fold_2_to_17	559 (-12)	-2.10%
hashmap	19,062 (-753)	-3.80%
brillig_rc_regression_6123	156 (-27)	-14.75%

[jfecher]

fold_2_to_17 1,168,465 (+21,730) +1.89%

I wonder what more is being executed here? When the only change is to issue fewer inc/decs, reference counts should in theory remain at 1 more often and just lead to more direct mutation.

[jfecher]

Approach taken here is incorrect. It'd lead to mutation if the same parameter was passed as a reference twice.