bench: add Ceno batched main sumcheck benchmark

[hero78119]

Problem

We need a benchmark shape closer to Ceno batched main sumcheck workloads, rather than only the smaller synthetic mixed-sum/product cases.

The existing devirgo_sumcheck benches are useful for focused frontload-vs-suffix comparisons, but they do not model a batched expression with many MLEs and heterogeneous product degrees across multiple chip-like groups.

Approach

Add a new Criterion bench target:

cargo bench -p sumcheck --bench ceno_batched_main

The benchmark builds a deterministic Ceno-like batched main sumcheck case with:

• multiple chip shapes
• many MLEs per chip
• many terms per chip
• product degrees mixed across 1, 2, 3, and 4
• deterministic RNG seed for stable generated MLE data

The default shape is:

scale=default
max_vars=20
mles=52
terms=150

Environment knobs:

Variable	Purpose
GKR_CENO_BENCH_SCALE	Selects tiny, default, large, or ceno-ish shape.
GKR_CENO_BENCH_MAX_VARS	Shifts the generated shape to a target max variable count.
GKR_CENO_BENCH_SAMPLES	Sets Criterion sample size.
GKR_CENO_BENCH_SUFFIX=1	Also runs suffix legacy mode for comparison.

The PR also fixes clippy nits in the new bench so workspace clippy stays clean.

Benchmark Result

Command:

GKR_CENO_BENCH_SUFFIX=1 GKR_CENO_BENCH_SAMPLES=10 cargo bench -p sumcheck --bench ceno_batched_main

Shape:

ceno_batched_main/default/max_vars_20/mles_52/terms_150

Mode	Mean
frontload	1.1678 s
suffix_legacy	8.9286 s

Frontload speedup vs suffix legacy: 7.65x.

Testing

Command	Result
cargo fmt	Pass
cargo make clippy	Pass
GKR_CENO_BENCH_SUFFIX=1 GKR_CENO_BENCH_SAMPLES=10 cargo bench -p sumcheck --bench ceno_batched_main	Pass; results above

Risk / Rollout

Risk	Mitigation
Benchmark runtime	Suffix mode is opt-in through GKR_CENO_BENCH_SUFFIX=1; default run benchmarks frontload only.
Shape mismatch	Exposes scale and max-vars env overrides for quick local tuning.
Flaky randomness	Uses deterministic RNG seed.
Clippy breakage	cargo make clippy passes after the bench additions.

This PR only adds benchmark code and does not affect prover/verifier runtime behavior.

Copilot Reviewer Directive (keep this section)

When Copilot reviews this PR, apply .github/copilot-instructions.md strictly.

• Perform review only; do not commit/push/propose code changes.
• Prefer inline comments on changed lines for each actionable finding.
• If inline comments are unavailable, use [severity] path:line (symbol) format.
• Prioritize soundness, performance, and architecture risks over style.
• Output order: findings by severity, then open questions, then brief summary.
• If PR description is empty or missing key context, report PR metadata: description as a finding.