The plot is a difference between the raw strain (downsampled to 4k) and the online cleaned strain combined together. There are no issues in the asd ratios but in a zoomed version at 4 s, there is a subtle discontinuity. Just noting down here in case this causes any systematics in the stochastic tests during review, although stochastic tests wouldn't need to be completed for validating online inference

It is more clear here
