[evlatests] Subarray 2 problems

[Walter Brisken]

I've transcribed a good portion of the idcaf log from Vivek's recent test 
showing odd labeling on subarray 2.  There is a common theme with every 
scan change: IDCAF gets 5 relevant documents from the executor:


Document start time  Document type            Document contents
-------------------  -------------            -----------------
54419.2290162037     Observation              140EVLTSTe.54419.217668495374
   = 05:29:47.000                              ORION KK 3333

54419.2290162037     VlaCorrelatorMode        5.0 sec mode 4 3333
   = 05:29:47.000

54419.2290162037     Observation              140EVLTSTe.54419.217668495374
   = 05:29:47.000                              0532+075 KK 2222

54419.2290162037     Observation              140EVLTSTe.54419.217668495374
   = 05:29:47.000                              ORION KK 3333

54419.2290162037     Observation              140EVLTSTv.54419.218175358794
   = 05:29:47.000                              ORION QQ 3333

Every scan change has the same set.  In the group is a new correlator mode 
(since that changed on most scan changes in this file), One new Observe 
document for subarray 1 (as expected), but _3_ Obs documents for subarray 
2, all with the same start time, but with different instructions.  IDCAF 
sometimes chooses incorrectly from the three choices.  This can explain 
everything Vivek complains about, I think.  In this case unexpected 
frequency labels get generated because the bandwidth code is not matched 
to the scan (making multiple freqids, where only 1 is expected).

Currently IDCAF cannot cope with conflicting documents.  I don't know of 
any way to change it such that correct document is always selected.


In my inspections, I see one case of a "first integration" problem after 
the integration time is changed.  I'll look into this since this is 
clearly an IDCAF issue.

-Walter