It appears that sometimes the reception of SNMP trap messages can somehow
get out of sync, and all too often this results in a scotty process which
apparently endlessly does a read() of 1 byte and gets EAGAIN, just to try
again. Instead, do some rudimentary check on the length of trap messages,
and try to re-sync the stream by finishing re-sync on first EAGAIN.
Bump PKGREVISION.
