[evlatests] unflagged records at the beginning of scans

[Gustaaf Van Moorsel]

The tests
---------

This message describes further testing of the problem that under
some conditions a number of records at the end of a scan are not
being flagged after the antennas have started moving.  These rec-
ords are then assigned to the *following* scan.

Last week I demonstrated that the problem is *not* specific to
fast switching: it showed up both in true fast switching and in
the fast switching emulating mode.

All those tests were done with 0.4 sec integration time.  New
tests expand this to 0.4, 1 2/3, and 3 1/3 seconds integration time.


Results
-------

At 0.4 sec integration time, the problem occurs at *every* source
change.   It can involve one or two integrations, never more than
two, never less than one.  At 1 2/3 and 3 1/3 seconds integration
time the problem was *never* seen in 28 source changes (14 fast
switching, 14 emulated fast switching).


Conclusion
----------

The current system appears to correctly flag all data after the
antennas have started moving for all integration times of 1 2/3
seconds and longer.  Of course this is no proof that the problem
is gone completely, and we should remain on the lookout for any
renewed occurrence.

Gustaaf