Server.Internal.ClientPing queue can grow very large

[scudette]

At a scale of around 50-100k endpoints we find that this queue can grow too large.

The queue is used to send client ping updates to the master (last time the client was seen etc) and as the number of clients grow this can saturate the replication service. The journal files are used as back pressure log files so they may eventually be delivered but this can grow to a very large size in high load scenarios.

We need to implement a few strategies here:

1. Rethink the volume of ping update messages to the server, can we reduce them?
2. Implement a maximum size of journal files so we can drop messages if the transmission rate is too slow for the message volume.

[scudette]

Fixed by #1700 should be available in 0.6.4-rc3