We currently already track validation failures through AFs. It could be valuable to have a dashboard to track these failures by failure type, since it'd give us a single place to go to see what failures we're generally seeing and to observe the rate at which these failures are coming in.

This slightly different from other fleet-wide metrics since it's not a value that's emitted by a live cluster and instead would come from fleet validation; if the existing AF infrastructure already gives us enough insight this ticket can be closed out.