This improves both scaling and fairness in the case where a single
track is being sent to large numbers of receivers. We avoid
communication overhead by growing the number of threads as the
square root of the number of receivers.
We used to send all of the packets corresponding to a received packet
as a single burst, which led to persistent packet drops for clients
at the end of the queue. Pace the writer, which introduces up to 1ms
extra jitter.