Add failsafe for exit if we drop events

[haesbaert]

If we drop events, especially exit events, that process will never be pruned from the cache.
To fix this we need to add a failsafe that triggers when we miss events but is also bound by time, say you lost X events, so somewhere in the next seconds/minutes we manually check if every process is still alive, probably by scraping proc.

Not labeling this as a bug as losing events is very very very hard.