starknet_transaction_prover: Prometheus /metrics endpoint with build_info

[avi-starkware]

Adds MetricsLayer that short-circuits GET /metrics with a Prometheus
text-format scrape, and an install_exporter helper that registers the
global recorder and emits prover_build_info with version + git_sha
labels. Wired alongside HealthLayer so the scrape path bypasses the
JSON-RPC parser.

Co-Authored-By: Claude Opus 4.7 (1M context) noreply@anthropic.com

[cursor]

PR Summary

Low Risk
Observability-only HTTP middleware and counters; no changes to proving logic, auth, or transaction handling beyond incrementing a metric on existing capacity rejections.

Overview
Adds Prometheus observability to the Starknet transaction prover: a GET /metrics scrape path and startup metrics registration, wired like the existing health endpoint.

MetricsLayer short-circuits GET /metrics in the HTTP/TLS tower stack (next to HealthLayer, inside request logging) so scrapes never hit the JSON-RPC parser. install_exporter installs the global Prometheus recorder at startup (before bind) and sets prover_build_info (version + git_sha labels) plus pre-registers prover_concurrency_rejected_total.

When prove_transaction rejects a request because the concurrency semaphore is full, the implementation now increments prover_concurrency_rejected_total. Dependencies metrics and metrics-exporter-prometheus are added; a unit test checks the scrape body includes build info.

^{Reviewed by Cursor Bugbot for commit 5eb413f. Bugbot is set up for automated code reviews on this repo. Configure here.}

[avi-starkware]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• starknet_transaction_prover: origin-level breadcrumbs for validation errors #14172
• starknet_transaction_prover: /health returns 503 when service is saturated #14171
• starknet_transaction_prover: bump prover_panics_total on panic hook fire #14170
• starknet_transaction_prover: HTTP request count + latency + in-flight metrics #14169
• starknet_transaction_prover: proving-job duration + outcome metrics #14168
• starknet_transaction_prover: Prometheus /metrics endpoint with build_info #14167 👈 (View in Graphite)
• starknet_transaction_prover: global panic hook + graceful SIGTERM shutdown #14166
• starknet_transaction_prover: per-request structured log with request-id #14165
• starknet_transaction_prover: redact URL credentials in logs and add startup banner #14164
• starknet_transaction_prover: add /health endpoint via HealthLayer #14163
• starknet_transaction_prover: add --log-format text|json CLI flag #14162
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[reviewable-StarkWare]

This change is 

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, have a team admin enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 9c4a5a0. Configure here.}

[cursor]

poll_ready delegates to inner but call may skip it

Medium Severity

MetricsService::poll_ready delegates to self.inner.poll_ready(cx), but call short-circuits for GET /metrics without ever invoking self.inner.call(). This consumes the inner service's readiness without a matching call, which can leak permits or slots in any inner service with stateful backpressure (e.g., Buffer, ConcurrencyLimit). The existing HealthService already handles this correctly by returning Poll::Ready(Ok(())) directly and not delegating to the inner service.

 

^{Reviewed by Cursor Bugbot for commit 9c4a5a0. Configure here.}