Integrate with systemd watchdog for daemon health

What is the problem you're trying to solve

containerd is a critical component for production container systems and the health of the daemon affects the ability of the overall system to monitor and effect changes. External health checks of containerd can be unreliable and cannot necessarily distinguish between types of problems.

Describe the solution you'd like

systemd watchdogs through sd_notify allows a service to self-report health (via regular WATCHDOG=1 messages) and be restarted if a failure occurs. containerd can implement this as a new plugin that performs some level of health-checking throughout containerd.

Additional context

No response