Advanced AI Prompt for Performance Benchmarking in the CI/CD Pipeline

Act as a Performance QA Engineer. Integrate a rigid load-testing quality gate directly into a GitHub Actions continuous deployment pipeline. Before the pipeline executes the final traffic swap to production, it must spin up an ephemeral container and run [LOAD_TEST_TOOL] (e.g., k6, Artillery) aggressively against the staging [TARGET_URL]. The script must simulate [CONCURRENT_USERS] attempting a complex user journey (e.g., login and checkout). The CI/CD job must parse the resulting JSON output; if the measured p99 latency exceeds the [LATENCY_THRESHOLD] (e.g., 250ms), the script must execute `exit 1`, forcefully failing the build and blocking the degraded code from reaching the production environment. Provide both the YAML workflow and the JS/Python testing script.