ci: Compare commit-comment benches by rc group

[samuelburnham]

• Enables the LURK_BENCH_OUTPUT="commit-comment" env var option, which configures the benchmark metadata for criterion-table formatting.
• Moves the reduction count out of the BenchmarkId name and into the BenchmarkGroup, which separates bench run tables by rc in criterion-table.

[huitseeker]

One nit: the fibo_prove / fibonacci_prove function naming doesn't hint at the benchmark organization. I'd rename fibo_prove to fibonacci_prove and call the old fibonacci_prove as fibonacci_prove_with_params or stg.

Differential CPU bench output for `LURK_RC="100,600" LURK_BENCH_OUTPUT="commit-comment"`

Benchmarks

Table of Contents

• Benchmark Results
  • LEM Prove - rc=100
  • LEM Prove - rc=600

Benchmark Results

LEM Prove - rc=100

	branch=ci-bench-comment-context	branch=ci-bench-comment-context2
num-100	12.50 s (✅ 1.00x)	12.39 s (✅ 1.01x faster)
num-200	24.50 s (✅ 1.00x)	24.46 s (✅ 1.00x faster)

LEM Prove - rc=600

	branch=ci-bench-comment-context	branch=ci-bench-comment-context2
num-100	9.69 s (✅ 1.00x)	9.59 s (✅ 1.01x faster)
num-200	21.57 s (✅ 1.00x)	20.34 s (✅ 1.06x faster)

---

Made with criterion-table

[huitseeker]

Thanks a bunch!