Fix readiness probe for containers without HEALTHCHECK

[yonghaoy]

Summary

• Fix probe-proxy-readiness.sh to use a Go template conditional ({{if .State.Health}}) instead of relying on docker inspect exit code to detect containers without a HEALTHCHECK
• This fixes non-SAS apps (jupyter, vscode, etc.) getting stuck in "Starting" state indefinitely

Problem

PR #415 added health check awareness to the readiness probe, but the Go template {{.State.Health.Status}} behaves differently across Docker versions when .State.Health is nil (no HEALTHCHECK configured). Some Docker versions return an empty string with exit code 0, bypassing the || echo "none" fallback. This causes APP_HEALTH="", which matches neither "healthy" nor "none", so the readiness probe fails forever.

Fix

-APP_HEALTH="$(docker inspect --format='{{.State.Health.Status}}' application-server 2>/dev/null || echo "none")"
+APP_HEALTH="$(docker inspect --format='{{if .State.Health}}{{.State.Health.Status}}{{else}}none{{end}}' application-server 2>/dev/null || echo "none")"

The Go template conditional handles nil .State.Health directly inside the template, making it work reliably regardless of Docker version.

Test plan

• Verify non-SAS apps (jupyter, vscode) transition to "Running" state correctly
• Verify SAS app still waits for healthy status before marking COMPLETE
• Recreate stuck VMs in vwb-dev-informal-olive-8524 after merge

🤖 Generated with Claude Code

[yonghaoy]

Closing in favor of #416 which correctly identified the root cause: docker inspect outputs an empty line to stdout before failing on containers without HEALTHCHECK, resulting in APP_HEALTH="\nnone" (leading newline). PR #416's | xargs approach correctly strips this. My Go template conditional fix also works but #416 was first and correct.