Node status inaccurate and inconsistencies

[jgresham]

Feature details

Sometimes when a client crashes, the client is not accurately marked as stopped or error. Additionally, the node status does not change from "running" to an error or stopped state.

Currently node crashes are detected by subscribing to podman events however if sometimes the events are not parsed properly or podman chould crash. We should additionally run checks every so often to monitor the status of nodes.