[evlatests] YET MORE on scan boundaries

[Michael Rupen]

I ran two quick tests using LST and UT scan boundaries, with 30sec
scans but asking the correlator backend to average 4 one sec LTA dumps to
give 4sec records out of the CBE.  The goal was to check that the CBE
does the right thing with scans which are not integral numbers of CBE
dumps, but *are* integral numbers of LTA dumps.

Yes, it does the right thing.  Note that in this case one can wind up
with 30sec LST scans which give only 6 four-sec CBE dumps, i.e., you
can lose up to two integrations (records) per (sub)scan.

Onwards --

            Michael