[evlatests] Scan change stuff

[Barry Clark]

I ran a file which cycles around three slightly different frequencies:
4616.00, 4616.05, 4616.15.  On scan changes, fringes vanish for a while
on the transitions 4616.05->4616.15 and 4616.15->4616.00, but not on
the transition 4616.00->4616.05.  This happens on all baselines.

They are usually gone for ten seconds, but sometimes twenty.

Some sort of shenanigans either in the Executor or Station Board
software.

Dataset SOFTWARE.tlo.evla.55131.67012083333 in the unlikely case
you care.