Enrich supervisor.status with next_run (epoch ms) on each schedule entry, and observe.status with running.duration_seconds and per-handler max_age_seconds for the health dashboard.