[evlatests] integer zeros banished

[Ken Sowinski]

On Wed, 7 Nov 2012, Rick Perley wrote:

>    We ran a 'PDif Compression' test last night.  In reviewing the data,
> I find that large fractions of some of the scans contain 'integer zeros'
> (and hence are removed by BDF2AIPS).  In column listings, some baselines
> are fully present, but about half the baselines have only ~25% of the
> data.

There was strong evidence that this was casued by a timing error
in dumptrig generation which occurs only rarely.  Bruce has repaired
this and we have rebooted all the station boards, so the problem should
no longer be seen.

This is the problem that in followup emails was called the Swiss
cheese effect.  Any examples of this from today on should be
brought to our attention.