[evlatests] Software time May 10

[Barry Clark]

Moderately disastrous.

Shortly after conversion to the Modcomp-free system, the CMP failed.
Not clear whether this was spontaneous, or something festering since
the planned power outage the previous day.  Very slow to come up again
after being rebooted (many minutes).  Hichem blames network problems....

Changes made last week to fix subarraying turn out to have serious
side effects on observation that change correlator mode or integration
time - often requires two or three source changes to get things straightened
out (if no other mode change occurs before then).  A touchy and difficult
area to work on, so an instant fix is improbable.

I tried using the CMP generated flags for VLA antenna off-source and 
subreflector positioned.  This resulted in all VLA antennas being 
flagged (CMP changes are needed to make these work better), and it
turned out to be difficult to get them unflagged again (the bit got
stuck in the software here and there and wouldn't go away).