Signal service readiness and enable the watchdog with Type=notify and WatchdogSec

Documented steps

1. Use Type=notify so dependent units wait until the app says it is ready, and WatchdogSec so a hung app is restarted: [Service] Type=notify WatchdogSec=30 Restart=on-watchdog ExecStart=/usr/local/bin/myapp NotifyAccess=main
2. In the app, after initialization completes, send READY=1 over the socket in $NOTIFY_SOCKET. C: `sd_notify(0, "READY=1");` Python (no deps): connect an AF_UNIX SOCK_DGRAM socket to os.environ['NOTIFY_SOCKET'] (prefix \0 if it starts with '@') and send b'READY=1'. Rust/Go have sd-notify crates/modules.
3. For the watchdog, read WATCHDOG_USEC from the environment and send `WATCHDOG=1` at intervals of at most half that value: `sd_notify(0, "WATCHDOG=1")` in the main loop.
4. If the gap between two WATCHDOG=1 messages exceeds WatchdogSec, systemd marks the service failed and kills it with SIGABRT (giving a core dump for diagnosis); Restart=on-watchdog (or on-abnormal/always) then restarts it.
5. Optional richer signals: `STATUS=...` free-text shown in systemctl status, `RELOADING=1`/`READY=1` around reloads, `STOPPING=1` on shutdown. systemd 253+ supports Type=notify-reload for reload handling via signal + notifications.
6. Verify: `systemctl start myapp` should block until READY=1 arrives; `systemctl show myapp -p WatchdogTimestamp,NotifyAccess,StatusText` confirms pings and status.