I have found that when the rotation rate goes slower than about 2 minutes per revolution that the sun sensors are no longer a reliable indicator for detecting the rotation rate. If you think about the collection/reporting scheme of the sun sensors, it makes sense. It is related to Nyquest but also because the data is collected for one minute and not collected for one minute. Also when the rotation rate got really slow, on the order of 8.5 minutes per revolution, the bird in all likelihood does not complete one revolution while within range to collect the high resolution data directly unless you just happen to be lucky.
I have been tracking the rotation rate from the beginning and get the same result independently as graphed above. Today the rate was 5.33 minutes per revolution.
So far I have come up with three differently hypotheses as to what is going on. Of course this is purely a mind game, I doubt if the hypotheses will ever to be tested and it is just going to take more data to refine the speculation.
One observation is that the temperature stability and distribution (compared between illumination/shadowed) are somehow related to the rotation rate even with the same illumination duty cycle. This bird just keeps on giving.