Load and stress checks
15 procedures covering requests per second and p95 latency with k6 or similar, memory growth under load, connection limits, queue depth, autoscaling triggers. Out of scope here: capacity planning, distributed load generation setup, cloud cost analysis.