The receiving side also switches timeout handling from the receiver to
the generator, which obviates the need for the sender to send any
keep-alive messages at all (for protocol 31 and beyond).  Given this
setup, all keep-alive messages are now sent as empty MSG_DATA messages,
with MSG_NOOP messages only being understood and (when necessary) acted
upon to forward a keep-alive event to an older receiver.  This is both
safer and more compatible with older versions.

MSG_NOOP is not available (is both safer and supports older rsyncs).

and no one is reading (i.e. do advantageous reading in perform_io()).

Added some debug output for MSG_ERROR_EXIT messages.

(Patch from Matt.)

all the time (we only waste from 1-3 bytes some of the time).

Make sure a write error drains any messages in the input buffer.

(while keeping it backward compatible). Improve --bwlimit docs.