[evlatests] more on bad subbands

[Rick Perley]

    I noted earlier that the LCP spectra usually have a bad subband in 
subband#6 (where I start counting at 1). 

    New spectra taken this afternoon  by Ken allow me to be more specific. 

    The bad subbands are seen again, always in LCP, and usually in 
subband #6.  But it is not on all baselines -- indeed, it is bad on only 
two baselines:  12 * 27 and 12* 28.  I have not yet checked all bands 
(there are 49 tunings, so this may take a while ...).