[evlatests] modcomp-free tests

[Gustaaf van Moorsel]

I ran three consecutive tests yesterday, each 30 minutes in duration:

1 - spectral line, regular C-band (H2CO), in order to test doppler
     tracking for VLA and EVLA antennas
2 - spectral line, extended C-band (methanol), in order to see how
     the new system deals with 'no change'
3 - continuum, regular C-band, in order to test changes Walter made
     to mosaicking

Disclaimer: some of the problems noted below may gave to do with the
IAT midnight problem.  Also, Ken warns me that 512 channels and 3
seconds integration may have been too much for the system to handle.

First results:

1 - Since I was not sure what program codes to use in FILLM, I filled
     with a blank program code, and specified the exact time range from
     the operator log.  To my surprise this caused FILLM to load 35 seconds
     of the previous X-band observation, program code PEVLC_.
     From the log:
     26 Jun 23:52:00 Starting program GVANTST
     From LISTR:
     1058+015   0/23:52:00 - 0/23:52:35

2 - The first 30 minutes, with program code W3H2CO, resulted in an AIPS
     file with program code AW3H2.  I assume this is according to the new
     naming scheme Walter put in

3 - The first part of the second 30 minutes, with program code W3OH, was
     loaded in the same file as the first, but with a different FREQID.
     This is the part of the second run that uses the same correlator mode
     as the first run, albeit with a vastly different frequency.  I'd have
     thought that a different program code would result in a new file in-
     stead of appending.  This may be an AIPS thing.

4 - The second part of the second run, with a much wider bandwidth (12.5)
     but still spectral line, somehow ended up as part of the file in the
     third run, a continuum mosaicking run at regular C-band.  So the third
     file created by FILLM consists of:
     a - the second part of the second run (spectral line)
     b - the third run (mosaicking, continuum)
     This file is labeled 'CBAND', in other words, a continuum file, and
     AIPS gives it program code AW30HB, which is not like the code of the
     second run (W3OH), or the third run (GVMMSF).

5 - The naming of the mosaicking samples continues to be different from
     the old system.  It insists on naming every pointing 1331+305, upon
     which AIPS adds modifiers when it sees that the positions do not
     agree.  A typical name is '1331+305     04'.

6 - When observing two consecutive and identical mosaicking scans, when
     starting the second scan, the system no longer continues the original
     sequence but goes back, as it should.  But it appears it does not go
     back to exactly the same position as the start of the first scan.
     Instead, it starts at the position of the *third* sample in the
     first scan.