argus instability problems

[Carter Bullard]

Gentle people,
I have new code on the server that attempts to address a number
of instability problems that were in the last release, primarily on
64-bit architectures.  I working the problem, we have improved
the threads model for argus and the clients, and so the code is
much better structured and will perform much better under load,
especially when there are multiple processors.

In particular this version should correct the "Zero length" error,
which did appear to be an argus() bug, rather than a radium() bug,
performance problems with radium(), where it would go idle
for many seconds at a time, and better handling of reliable connections
to multiple remotes for all the ra* programs.

Please give this a run, and hopefully this will be the final release
candidates for argus-3.0.0.

Carter