[evlatests] Last 10 seconds issue

[Rick Perley]

    It seems that it is always true that at precisely 10 seconds before 
the end of a scan, when the following scan is at a different frequency, 
the VLA and EVLA become incoherent.  The VLA antennas are always flagged 
out during the interval, the EVLA antennas are not. 

However, it is clear from the data that the VLA-VLA correlations 
correctly continue through the interval (so the flagging is incorrect), 
while the EVLA-EVLA correlations suddenly change in both ampliutude and 
phase at the 10-second mark (so they should be flagged). 

Further evidence that the problem is on the EVLA side comes from the 
Tsys records.  There is no change whatever for all the Tsys values for 
the VLA antennas, whereas the EVLA antennas abruptly change Tsys, as if 
they had been commanded to a new frequency.