This change adds a "watchdog notifier" utility which tracks the health of request handlers and notifies an external systemd watchdog if everything looks good. This allows us to cause a termination if any request handlers get hung.