CH credential.

When set, randomizes the order of nodes before making pclasses. This
has the effect of putting different nodes at the 'front' of the pclass,
The idea is to avoid 'hot spots' of nodes that get over-used simply
because they happen to appear earlier in the .ptop file.

NB: because STL seems to use a different source of randomness, using
this with '-s' to set a seed does *not* make assign deterministic.

table instead.

iface, leave the lannode side alone. This is probably the cause
of infrequent swapmod probems with experiments that have lans.

Inster node_auxtypes entry for the capacity.
Minor bug fix.

Also add a test script helper for stitching.

Ryan fixed this before, but apparently didn't check it in.
This commit also revisits the setting of the "max" values for buffering
limits and distinguishes the default values from the max.

The geniuser is now over 16 groups, so try to figure out what project
the action is going to take place, and use just those groups. This is
temporary cause Mike says FreeBSD8 8 pushes the limit up to 1024.
Yippie! When we upgrade, just revert this crap.

tree mode.

a little bit to avoid a race at the very end of time, when trying to
unregister a sliver from the SA.

... we don't do it for any other error, and it's best not to inject
user input back into the HTML page; presumably totally bogus strings
could evaluate to 0, and thus look 'too small'

It's best not to echo any form input back into the page without
cleaning it first, and in this case, it's not that important to
the message.

Also fix the conditional that was causing it to not
get printed.