Fix /health robustness

[casey-brooks]

Summary

• make Docker runner status optional in the health controller and ensure a deterministic 200 response with cache-control
• align platform-server helm probes with /health
• add controller tests for injected, missing, and throwing docker-runner status service

Testing

• /root/.nix-profile/bin/pnpm lint
• /root/.nix-profile/bin/pnpm --filter @agyn/platform-server exec vitest run tests/infra/health/health.controller.test.ts

Curl evidence

curl -i http://127.0.0.1:3310/health

HTTP/1.1 200 OK
cache-control: no-store
content-type: application/json; charset=utf-8
content-length: 211
Date: Tue, 03 Mar 2026 22:11:34 GMT
Connection: keep-alive
Keep-Alive: timeout=72

{"status":"ok","timestamp":"2026-03-03T22:11:34.937Z","checks":{"dockerRunner":{"status":"skipped","optional":true,"error":{"name":"DependencyUnavailable","message":"DockerRunnerStatusService not registered"}}}}

Fixes #1364

[casey-brooks]

Test & Lint Summary

• /root/.nix-profile/bin/pnpm lint
• /root/.nix-profile/bin/pnpm --filter @agyn/platform-server exec vitest run tests/infra/health/health.controller.test.ts

Results: lint passed with no errors; tests passed (3 total).