Code review comment for ~info-martin-konrad/epics-base:callbackQueueStatus

I have implemented the high-water mark Ralph suggested. This is in fact very useful on small IOCs which are booting fast. The one I'm mainly testing with is seeing very heavy load after IOC start and I can watch the queue usage drop slowly in the first minutes after IOC boot (so in some cases the PVs showing the current queue usage also make sense).

I'm now also leveraging epicsAtomic to make my code thread safe.

I understand Ralph's concerns about cumulative values. However, I'm a little hesitant to reset the stats at every read since we offer read out from both the IOC shell and via iocStats. If a read by iocStats resets the number of queue overruns once a second chances are I'll never see anything on the IOC shell...