feat(bench): add bounded external HTTP load

[chubes4]

Summary

• add a bounded external-http-load benchmark step that targets the active runtime preview through host-side HTTP
• preserve one runtime while collecting status, concurrency, latency, diagnostics, and explicit transport provenance
• merge measured load evidence into the existing wp-codebox/bench-results/v1 scenario and retain the public workload definition
• fail closed on invalid bounds, cross-origin targets, transport failures, incomplete execution, and unexpected statuses

Closes #2413.

Verification

• npm run test:external-http-load-integration
• npm run test:bench-command-step-behavior
• npx tsx tests/benchmark-contracts.test.ts
• npx tsx tests/runtime-wp-cli-bridge.test.ts
• git diff --check

AI assistance disclosure

OpenAI GPT-5.6 Sol via OpenCode was used to inspect the benchmark architecture, implement the host-side load primitive and result projection, add deterministic and WordPress integration coverage, and run the verification commands. The operator reviewed the implementation direction and resulting evidence.