fix: apply default healthcheck intervals for readycheck sidecars

[dvush]

When ReadyCheck fields (Interval, Timeout, etc.) are zero-valued (e.g. from YAML recipes), Docker interprets "0s" as "use defaults" which means a 30s polling interval. This caused services like ClickHouse to appear to take ~30s to start when they were actually ready in ~2s.

Apply sensible defaults (1s interval, 10s timeout, 30 retries).

[claude]

Claude finished @dvush's task in 57s —— View job

---

PR Review: fix: apply default healthcheck intervals for readycheck sidecars

• Read and analyze the changes
• Provide review feedback

LGTM. The change is well-scoped and solves the right problem. YAML recipes create ReadyCheck{QueryURL: ...} with zero-valued timing fields, which get serialized as "0s" into docker-compose healthcheck config — causing Docker to fall back to its 30s default interval.

One minor observation, not blocking:

The defaults are applied only in the sidecar code path inside Validate(). If a non-QueryURL ReadyCheck is ever constructed with zero-valued fields (e.g. from a future YAML test field), it would hit the same "0s" issue at local_runner.go:676-681. You could consider applying defaults in WithReady() instead to cover all paths, but that's a broader change and not needed for the current use cases since all Test-based checks in components.go already set explicit values.

---