[evlatests] Flagging oddities, Nov 02.

[Ken Sowinski]

On Fri, 2 Nov 2007, Vivek Dhawan wrote:

> There were low amplitude (off source) points in near-identical
> pattern on every baseline. The pattern was that bad points, that
> should have been flagged, occurred in bursts for several scans
> in a row, and then not at all for 10's of minutes. Then after
> the first 3 hours there were no bad points at all.

Unused startup time before the baseline time was used to
measure collimations.  The data was useless because the
pointing determinations were nonsense in a way that could
be consistent with badly flagged data.  A similar run over
lunch today provided usable data.

Something funny was happening last night.