Remove reset instruction

[Schaeff]

Remove the _reset instruction in the dispatcher.
When turning asm into pil, we have a dispatcher which first resets all registers and jumps to the correct part of the ROM. As a result, the write registers are unconstrained on the first row.

After this PR, registers are reset after calling return, just like for the pc. This removes the need for the _reset instruction, as long as we constrain the write registers to be 0 on the first row.

Here are the tradeoffs of this change:

• pros
  • simplify the dispatcher, as now the pc and write registers are treated more similarly and we remove an instruction
  • fully constrain the write registers, including on the first row (note: this could also be achieved by constraining them to an arbitrary value on the first row)
  • save a main column instr__reset
  • save a ROM column p_instr__reset (negligible)
  • save a ROM row (negligible)
• cons
  • adds a witness column per write register: because of the initialisation to 0 on the first row, the register update is now higher degree. Maybe this could be an intermediate column?

[github-actions]

⚠️ Performance Alert ⚠️

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

Benchmark suite	Current: fec80a8	Previous: 0bdbbd2	Ratio
jit-benchmark/sqrt_879882356	3616 ns/iter (± 4)	2735 ns/iter (± 3)	1.32
jit-benchmark/sqrt_1882356	2861 ns/iter (± 12)	2165 ns/iter (± 4)	1.32
jit-benchmark/sqrt_1187956	2857 ns/iter (± 6)	2126 ns/iter (± 4)	1.34
jit-benchmark/sqrt_56	1687 ns/iter (± 2)	1261 ns/iter (± 1)	1.34

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

[Schaeff]

I think the benchmark alert is a false positive because #2264 is out of sync with main. The regression was introduced in 8dd0a00

[leonardoalt]

lgtm, but maybe you want someone else to also take a look?

[leonardoalt]

has a bunch of conflicts now =/

[lvella]

Did you base this PR off some other PR that was merged? Then you need to merge main back into the branch of that base PR, then merge that branch into the branch of this PR, and all the conflicts and spurious commits will magically disappear...