To study the variation over time of the jitter peaks in the sensitivity, I computed the band-limited RMS of CAL-DELTAL_EXTERNAL between 300 and 400 Hz, for all the reasonably long lock stretches of the last three weeks. The attached plot shows the result: blue dots are the values over one second periods, while the orange traces are smoothed version over 10 minutes.

I was hoping to see some variation or trend during each lock, but apparently the amplitude of the peaks are quite constant, and it doesn't change much over each single lock. There are few exceptions: a couple of locks show very low values, but looking at them in more detail it looks like the entire sentivity was scled down, so I'm not sure if they are meaningful. Moreover, there are a couple of cases where the amplitude of the peaks suddendly decrease or increase, for example around GPS 1124092417 (Aug 20 2015 07:50 UTC) and 1124863017 (Aug 29 2015 05:55 UTC).