[evlatests] Offset between data and flags?

[Rick Perley]

    I ran a test script last night, observing three strong sources 
sequentially.  The sources are sufficiently separated that about 1 
minute of slew time was needed for each source change to get on source.  
I used 3 seconds on-line averaging.  Examining the data carefully 
revealed an interesting offset:

    For every one of the 20 or so scans (no exceptions!), the on-line 
flagging was 3 seconds off, in the sense that there was one good 
3-second record at the beginning that was (probably) erroneously 
flagged.  Admittedly, this is a pretty small issue.  But is it possible 
we have a timing offset between the flagger and the data, when time 
averaging is applied?