It is easier to understand peerTracker health if we make the observe callback directly look at the length of the map at the time of observation rather than observing the ints that are on the exchangeMetrics struct as it's not clear when the peerTracker has X many tracked vs disconnected peers.
Let's instead take the length of the maps directly (rlocking of course), but this will make the metric more understandable.
It is easier to understand peerTracker health if we make the observe callback directly look at the length of the map at the time of observation rather than observing the ints that are on the
exchangeMetricsstruct as it's not clear when the peerTracker has X many tracked vs disconnected peers.Let's instead take the length of the maps directly (rlocking of course), but this will make the metric more understandable.
cc @distractedm1nd