Harden develop CD deploy step with retry on transient health-check failure

[Copilot]

The build-and-deploy GitHub Actions job was failing in the remote deploy phase even though build and artifact transfer succeeded. Logs show intermittent health-check timeout behavior ending with deploy script exit before Nginx switch.

• What changed

  • Updated .github/workflows/backend-cd-develop.yml in Run deploy script on develop server.
  • Replaced single-shot deploy invocation with a bounded retry loop (2 attempts, 30s backoff).
  • Preserved fail-fast semantics after the final attempt (exit 1) so real persistent failures still fail the job.

• Why this addresses the failure mode

  • The failing run pattern indicates transient startup/health-check instability rather than deterministic build or transfer errors.
  • A single controlled retry absorbs brief warm-up variance without weakening failure signaling.

• Snippet (workflow change)

  script: |
    for attempt in 1 2; do
      echo "> Deploy attempt ${attempt}/2"
      if ${{ secrets.DEVELOP_DEPLOY_SCRIPT_PATH }}; then
        echo "> Deploy succeeded on attempt ${attempt}"
        exit 0
      fi
      if [ "${attempt}" -lt 2 ]; then
        echo "> Deploy script failed. Retrying in 30 seconds..."
        sleep 30
      fi
    done
    echo "> Deploy failed after retries."
    exit 1