Prometheus and PeerLastSeen out of sync

[t-lin]

After a long while, Prometheus seems to have some metrics for peers with 0 RTT that do not exist in ping-monitor's PeerLastSeen data structure. Similarly, there are occasional error messages that appear in ping-monitor trying to delete gauges (from PeerLastSeen's expiry mechanism) that do no exist in Prometheus.

This can be seen by calling hl-cli list periodically at an interval over a long period of time. A few metrics w/ 0 RTT should appear in the dashboard but do not exist within the PeerLastSeen structure.

Attempted to wrap all access to Prometheus client code in ping-monitor using mutex, but that doesn't seem to fix it.

[t-lin]

Easiest "solution" (more of a work-around) is to simply have a seperate goroutine sync the two every x minutes.