[evlatests] mosaicking, modcomp-free

[Gustaaf van Moorsel]

I ran the following test twice:

1 - 1 scan on a strong calibrator
2 - 1 mosaicking scan on that same calibrator, mode RSL
3 - 1 mosaicking scan on that same calibrator, regular mode

first in Modcomp-free mode, then in Modcomp mode.

Results for the Modcomp-free mode:

1 - In the first scan, only useless (low amplitude but unflagged) data
     on all baselines involving a VLA antenna.  This in spite of the fact
     that the scan duration is 8 minutes
2 - In the first scan, for EVLA - EVLA baselines, there were useful data
     but it took 4 minutes to start producing them.  During the first 4
     minutes there were similar low-amplitude data as on VLA baselines
     during the whole scan
3 - Unlike under Modcomp control, no special source names are created for
     each of the pointings within a scan.  Aips will assign different names
     if it sees different pointing positions, but it cannot tell the source
     in the first (non-mosaic) scan apart from the first pointing on that
     source in the mosaic scan
4 - Whereas under Modcomp control scan 3 (mosaicking scan 2) starts on
     the calibrator, as it should, in Modcomp-free mode it continues
     moving away from the calibrator.  In other words, in Modcomp-free
     mode the beginning pointing of scan 3 is the final pointing of scan 2
5 - There are minor differences in the actual pointing positions among
     the samples in the mosaicking scan, suggesting a different algorithm
     is being used.  I don't know how serious this is, but it may be worth
     investigating.

Gustaaf