[evlatests] Software time March 22 (continued)

[Barry Clark]

I have set the delays for the Modcomp free system for X band, IFs
A and B.  The reference antenna, ea16, had such a large delay error
in IFs C and D that I don't regard the measured results as reliable,
so I didn't set those.  (Antennas 21, 24, and 17 were out of the array.)

I have concluded that the distinguishing feature of the script that
reliably fails is that it switches correlator mode every scan (among
1A, 1B, 1C, 1D).  That should be enough of a clue to diagnose the problem,
though I haven't yet.

There was an interesting incident in which the correlator controller was
in continuum mode and idcaf was expecting spectral line data.  The resulting 
data looked a little strange.

Changing subarrays doesn't work like I expected it to, but I have not
yet figured out how it does work.