This new setting limits the number of UDP queries we attempt to
handle after being woken up by the multiplexer and before returning
back to process other events.
Before this, we could end up trying to process queries after queries
and almost never return from handleNewUDPQuestion() to process new
events, meaning we could eventually end up never scheduling new
mthreads or handle responses from authoritative servers for a long
time, only sending responses for packetcache hits and creating new
mthreads.