chore: benchmark ultrahonk circuits

[ludamad]

Adds UltraHonk proving benchmarks for the base parity circuit at 8, 16, and 32 CPUs.

Changes

• yarn-project/ivc-integration/src/base_parity_inputs.test.ts: Generates bytecode + witness for the base parity circuit. No-ops unless BASE_PARITY_BENCH_DIR env var is set (only runs during build_bench).
• barretenberg/cpp/scripts/ci_benchmark_ultrahonk_circuits.sh: Runs bb prove --scheme ultra_honk --verifier_target noir-rollup, verifies the proof, and extracts 24+ component timings from the hierarchical breakdown. Outputs benchmarks.bench.json.
• yarn-project/end-to-end/bootstrap.sh: Integrates input generation into build_bench (with caching) and adds 3 bench_cmds entries for CPUs 8, 16, 32.
• .claude/skills/adding-benchmarks/SKILL.md: Documents the extra: "stacked:GROUP_NAME" convention for rendering component breakdowns as stacked charts on the dashboard.

Benchmark entries (per CPU count)

• ultrahonk-bench/parity_base/cpus-{N}/total_ms
• ultrahonk-bench/parity_base/cpus-{N}/memory_mb
• ultrahonk-bench/parity_base/cpus-{N}/proof_size_kb
• 24 component timings (OinkProver, sumcheck, commitments, etc.) with extra: "stacked:..." for grouped visualization

[AztecBot]

Flakey Tests

🤖 says: This CI run detected 1 tests that failed, but were tolerated due to a .test_patterns.yml entry.

\033FLAKED\033 (8;;http://ci.aztec-labs.com/23d2bf0087735350�23d2bf00877353508;;�):  yarn-project/end-to-end/scripts/run_test.sh simple src/e2e_epochs/epochs_l1_reorgs.test.ts (397s) (code: 1) group:e2e-p2p-epoch-flakes (\033ludamad\033: Merge branch 'next' into ad/test/benchmark-ultrahonk-circuits)