feat(perf): report SDK benchmarks as named OTLP log events

[Jackson Weber (JacksonWeber)]

Measure genuine SDK throughput and signed memory deltas, preserve raw artifacts, and generate named OTLP log results with explicitly configured export.

[Copilot]

Copilot review overview

🟡 Changes recommended

Unresolved critical and moderate findings affect benchmark reliability, artifact preservation, and export correctness.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 2

Open (2)

• MetricReader subclass lacks required onCollect implementation · New
• Symlink or hard-link output can overwrite raw artifact · New

What changed in this PR

Adds SDK performance benchmarking with signed memory measurements, raw artifact validation, and explicit OTLP log-event export.

Changes:

• Adds span, metric, and log throughput benchmarks.
• Validates and preserves raw benchmark artifacts.
• Generates and exports named OTLP log events.
• Documents benchmark workflows.

File	Summary
test/​integration/​performance-events.test.mjs	Tests benchmarking, validation, and export behavior.
perf/​report-results.mjs	Validates artifacts and creates OTLP events. Moderate finding: scenario identity fields can be relabeled.
perf/​memory-worker.mjs	Captures isolated memory measurements.
perf/​export-events.mjs	Writes and exports event payloads. Critical finding: aliased output paths can overwrite raw artifacts. Moderate finding: valid empty 2xx responses are rejected.
perf/​benchmark.mjs	Runs throughput and memory benchmarks.
perf/​benchmark-sdk.mjs	Configures isolated SDK workloads. Critical finding: the metric reader uses an abstract API incorrectly. Moderate finding: detector suppression may target the wrong dependency tree.
CONTRIBUTING.md	Documents benchmark and export workflows.
CHANGELOG.md	Records the new performance tooling.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

Copilot review overview

🔵 Needs a closer look

Scenario validation permits fabricated test and category identities in exported events.

Review effort: Balanced
Findings: None

Resolved since last review (2)

• Symlink or hard-link output can overwrite raw artifact
• MetricReader subclass lacks required onCollect implementation

Previously missed (1)

In code that hasn't changed since last review

Validate all canonical scenario identity fields

perf/​report-results.mjs:102

The validation only compares scenario names, so a raw artifact can change a scenario's test and category (and make the same change in its memory entry) and still emit records under fabricated identities. Compare all three canonical identity fields from scenarios, not just name, so the named events cannot be relabeled while passing validation.