LIVENESS only: is this process running? Railway restarts the container on this.

The schema signal moved to /readyz (#2609). Restarting cannot fix a schema mismatch, so reporting one here bought a crash-loop instead of a fix, while a container that is merely busy — or one waiting on migrations — looks dead. #2028's requirement (a failed schema must not be invisible) is preserved by /readyz, which deploy-api.sh checks after every deploy; it just stops being wired to the restart trigger.