prometheus: evicting old entries in top_domains/requesters metrics

[dmachard]

The following top metrics need to be updated to remove old top entries

• dnscollector_top_domains
• dnscollector_top_requesters
• ...

[v0rdych]

Hi, maybe it's worth to add new setting (measured in seconds) in configuration?
Which will be used in separate thread clearing all TopMap structures based on that setting?

I've tried just clear all TopMap entries when metrics are collected by external system and it also seems working solution at least when one system is collecting metrics. And this also works with that fact, that all "gaugeTop**" metrics seem to be not GAUGE, but COUNTER in nature.
So clearing TopMap structures in the moment of collecting entries and changing grafana dashboard's formulas to rate instead of increase gave me correct messages-per-second values of top-domain entries.
But still have no solution when metric collection is performed from multiple systems, so *TopMap structures are cleared too early.