Reason for change (UPDATE)
Following an analysis of the unbinned photons above 3 GeV, it appears to actually be a fixed offset
see this link
-As a follow up to the C&A meeting where I presented the results of pass6/pass7 boresight comparison, it appears that the alignment has changed by 60 arcseconds since the alignment was first determined during L&EO. In particular, the alignment seemed to change over the period from December to February the most significantly. I propose that one set of boresight alignment calibrations should be split up as follows:
...
Here is the plot from the C&A meeting:-
Test Procedure
The pointlike boresight alignment analysis will be run locally to verify the constants have been applied correctly.
...