when the timeout event was fired, it started another step in query
resolution, but there may have been pending events on previous handle
if the events on the previous handle completed, they would check the
memory of their task (which may have been non-existent) and possibly
double-close the handle

no need to scramble queries satisfied from cache

an average request from cache requires ~12k, an average iterative request 16-64k
this reduces allocation cost when there are now pools on the freelist

since the structures are ~ same size as I/O requests (udp_send_t being 60B larger than others), they all share the same freelist

the stats module keeps track of frequent iterative queries using
probabilistic sampling. the list includes subrequests.
it is useful for tracking over time, prefetching and anomaly reporting

enumerated metrics are always present (set to 0 if not used), this fixes
missing values when not measuring and provides a faster access to common
data

this can coalesce sends/writes in future versions of libuv

this fixes an issue with libuv sometimes crashing with mismatching number
of descriptors with REUSEPORT