Automated performance regression detection in ci pipeline

[clintjeff2]

Motivation

• Prevent unintended latency regressions on critical paths by enforcing a committed P99 baseline and an automated CI gate.
• Provide a reproducible, small JSON snapshot format so benchmarks, observability exports, and load tests can feed the same gate.
• Ensure the system meets the project's performance SLOs (hard SLO: 100ms P99, regression budget: 10%) before merges.

Description

• Add a committed baseline of critical-path P99s in .github/performance/baseline.json describing meter_heartbeat_ingest, usage_settlement, oracle_rate_lookup, and dashboard_usage_summary.
• Implement scripts/performance_regression_gate.py, a standalone Python CI gate that validates snapshot JSON, compares baseline vs current, applies a hard --max-p99-ms limit and a --regression-percent budget, prints pass/fail markers, and can emit a Markdown report.
• Wire the gate into the test-and-lint job in .github/workflows/ci.yml to run after cargo test and upload a performance-regression-report artifact.
• Add documentation docs/PERFORMANCE_REGRESSION_CI.md, update README.md to include the gate and local invocation, and add unit/CLI tests at tests/performance/test_performance_regression_gate.py.

Testing

• Ran the Python unit tests with python3 -m unittest tests.performance.test_performance_regression_gate, which passed.
• Executed the gate end-to-end with the committed baseline using python3 scripts/performance_regression_gate.py --baseline .github/performance/baseline.json --current .github/performance/baseline.json --max-p99-ms 100 --regression-percent 10 --report /tmp/perf.md, which succeeded and produced a Markdown report.
• Verified repository checks: git diff --check passed, cargo fmt --all -- --check passed, cargo test passed, and cargo clippy --all-targets --all-features -- -D warnings passed.

---

Closes #65