[codex] Cache monomorphic array guard feedback

[andrewtdiz]

Summary

Adds a small lock-free, direct-mapped fast cache for monomorphic array typed-feedback guard sites. The cache is seeded by the existing slow guard_observe path and only fast-passes when the current array observation exactly matches the cached feedback key and the runtime contract guard is valid.

Slow paths still update the typed-feedback registry, failures, megamorphic state, observations, fallback counters, and trace metadata. Trace snapshots merge cache fast-pass counts back into observed_count, per-site guard passes, and by-guard totals.

This is stacked on #5302.

Benchmark Evidence

Baseline at ed71efde8585 from #5302:

• Direct matrix runs: 1736ms, 1730ms, 1729ms, 1738ms, 1714ms; checksum always 41079519680
• quick.sh matrix: 1745ms / 28MB
• perf stat matrix: 28.04B instructions, 4.65B branches, 6.34B cycles
• Typed-feedback trace: 33,619,968 numeric array get guard passes and 65,536 set guard passes with 0 get/set failures

After this change:

• Direct matrix runs: 1239ms, 1258ms, 1223ms, 1247ms, 1226ms; checksum always 41079519680
• quick.sh matrix: 1228ms / 30MB
• perf stat matrix: 16.74B instructions, 3.09B branches, 4.49B cycles
• Trace still reports 33,619,968 numeric array get guard passes and 65,536 set guard passes with 0 get/set failures

Measured impact:

• Direct matrix median: 1730ms -> 1239ms, 28.4% faster
• quick.sh matrix: 1745ms -> 1228ms, 29.6% faster
• Matrix instructions: 28.04B -> 16.74B, 40.3% fewer
• Matrix branches: 4.65B -> 3.09B, 33.6% fewer
• compare.sh --quick --runs 3 --warn-only nested_loops median: 921ms -> 662ms, 28.1% faster

Verification

• cargo fmt --check
• git diff --check
• cargo test -p perry-runtime typed_feedback
• cargo test -p perry-codegen --test typed_feedback
• cargo test -p perry-codegen --test typed_shape_descriptors
• PERRY_BIN=target/release/perry python3 tests/test_typed_feedback_runtime_evidence.py
• tests/test_benchmark_output_verifier.sh
• cargo build --release
• target/release/perry compile --no-cache benchmarks/suite/16_matrix_multiply.ts -o /tmp/perry-matrix-array-guard-cache-final --quiet
• benchmarks/quick.sh
• benchmarks/compare.sh --quick --runs 3 --warn-only --json-out /tmp/perry-array-guard-cache-final-ed71efde8.json

Note: benchmarks/baseline.json is stale for this Linux environment, so compare was run with --warn-only and the PR body uses the captured local #5302 baseline.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 8109785b-88d0-4cc4-b12b-9b9412d43206

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch codex/perry-array-guard-cache-fastpath

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}