QC not capturing double dipping and/or lick graphs adequately

This mouse (780619, Avalon) was failed yesterday for bias:
https://qc.allenneuraldynamics.org/qc_app?id=ac84f963-cf6b-484f-9e65-0202f3037ce9&active_evaluation=2

Looking at the behavior graph in the GUI before I trained the mouse for Avalon, I noticed that the lick graphs look quite different. I don't think that the QC adequately captured the double dipping (licking left and right in quick succession) that this mouse struggles with. Looking for input/thoughts from others.

I do agree that the lickspout should have been moved in the session on 3/26 - when I trained this mouse today I was able to get him to stop double dipping, and switch according to probability changes much more effectively by priming with manual water and moving the lickspout.