test(bench): execute pinned Graphalytics fixtures through the public API

[DecisionNerd]

The Graphalytics adapter previously validated replayed output without proving execution through GraphForge. Add a bounded ga-tiny path through the public Rust facade, with pinned Graphalytics 1.0.5 identities and reference semantics. BFS, WCC, and SSSP execute and pass validation; PageRank, CDLP, and directed LCC emit explicit typed semantic incompatibilities, as required by #961. Live execution and historical static replay retain separate evidence identities.

Bind the native command to its compiled input identity, actual edge/reference hashes, all six algorithms, and pinned job parameters. Reject forged or incomplete inputs before emitting evidence, and reject out-of-range numeric results without saturating conversions.

Update the benchmark dependency check for live facade execution while preserving private benchmark packages and rejecting engine-internal bypasses, unrelated local paths, and symlink escapes. Existing certification/storage and generator/filesystem helpers remain limited to their exact packages and manifest paths.

Validation on the integrated tree:

• make -C benchmarks smoke: passed, including 368 Python tests, all Rust runner tests (10 Graphalytics tests), benchmark workspace Clippy, formatting, and the actual Cargo dependency graph.
• Real CLI tests cover pinned-input rejection and the three passes/three typed incompatibilities; this is not a six-algorithm compatibility claim.
• Eleven dependency-boundary regressions cover allowed and rejected paths.
• make pre-push-fast, focused all-targets Graphalytics Clippy, and final diff checks: passed.

Closes #961. #952 remains the broader M11 completion gate. Exact-head CI passed at 7621f3c551b735f3fec65b741c8cb2798577aacf, including authoritative Bazel Rust tests, native lifecycle execution, binding packaging, platform durability, concurrency matrix, and CI Gate. Squash merged; #961 is closed.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. 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: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Team

Run ID: d4393839-26cf-4bdb-8656-e8f82e19bb1c

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

Warning

Billing warning: we have not been able to collect payment for this subscription for more than 72 hours. Please update the payment method or pay any pending invoices in Billing to avoid service interruption.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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