[evlatests] Small frequency offsets

[Barry Clark]

My turn to join Rick in the mea culpa department.  The behavior I was
reporting earlier, of vanishing fringes, was a pilot error.  I had not
entered a fshift in the shifted LOs.  (Thanks Ken for telling me.)

Things look more reasonable today, but still not quite right.

There is a phase jump ten seconds into the thirty second scan,
rather than at scan boundary.  Somewhere there is a misunderstanding(s)
about what time things are happening.

May be a little more complicated than a ten second offset.  The first
scan is longer, and there is a phase jump 20 seconds before it ends.
This suggests that the phase jump may be 20 seconds before the end
of the scan, instead of 10 seconds after the beginning.

I'll try rerunning this with longer scans to clarify.