Implement external monitoring of public cluster #507
Description
Consider setting up a monitor that periodically makes a simple request to the public cluster (maybe every minute, or every 5 minutes), and if the request fails, there is a text message or email that is sent to the sliderule developers.
This could also be combined with (or implemented as) the monitoring functionality in Grafana. We have metrics on the number of container restarts, nodes registered, and discovery failures - could we put together some heuristics that determine when to alert the developers from those?