[evlatests] Subarrays

[Barry Clark]

Subarrays work, but they are extremely fragile, and behave oddly.  So
far I have just been collecting behaviors for later investigation.
Behaviors include:

It takes a while to get organized.  Data do not get sensible until after
at least one source change in each subarray, after they are started.  
(Data in both subarrays are nonsense until this occurs.)

Starting the subarray with fewer antennas first seems to work reliably; 
the one time I tried starting the larger first didn't work.

Starting a new script in either subarray causes it to stop working
altogether.

Amplitudes are lower in one case I looked at, in continuum mode in IFs
AC, for an EVLA antenna pair in the minority subarray.

Line mode subarrays seem to work, but it took even longer to get organized.
Data did not start flowing until after a correlator reconfiguration after
a scan change in the other subarray.  The amplitudes and delays look 
pretty nominal for these mode 1x observations.  (I had been expecting 
that the low amplitudes in continuum would have been a delay error.)