My own event a couple nights ago didn't completely turtle, but was heading on that trajectory pretty quick. Today was another near-0F commute with high teens on the return trip. I ran leafspy this time while driving to see if it would act up again today (it did not) and to watch cell balance and actual battery temps.
Think I'm seeing the problem now... Battery temps today were relatively low (3 segments on dash) and showing in the low-to-mid 30F's on leafspy, SOC was under 40%. The cell balance becomes very unbalanced (say in 50+ mV range) if the car is pulling higher power loads such as @ 3KW to heater and pulling a grade at expressway speeds. Under those conditions, certain cell pairs (#50 or 51, and a few adjacent cells) basically drop off the bottom of the leafspy voltage graph. Once the battery temps warmed slightly (@ 4 segments & in 40F's on leafspy) or the load dropped, the cell balance leveled out and looked very normal. In hindsight, the events from a couple night ago was along the same pattern. The SOC cratered when pulling a substantial long expressway grade and then recovered when I headed for the exit.