OK - so this is the third analysis - looking again at an example with wrists close together (Inga's arms are 80 degree in a full hang) to test the hypothesis that this discourages hanging and encourages arching.

Here is the raw data, chart and key:

https://imgur.com/a/GmzOo8p

I multiplied the scores in the first clip by 1.6 to account for the shorter duration.

1. One thing I notice about this is that the first two clips already diverge quite significantly. Even after only 5 minutes on the cross Inga is obviously tiring. During minutes 3, 4, and 5 she is arching for almost three times as much time as in minutes 1 and 2. Her patterns in general are not as clear, perhaps in part due to what I think is quite rapid exhaustion. I hypothesize that she is not as fit or strong as some other models.
2. That said, there are clear trends. She spends dramatically more time arched in the last six minutes, significantly less time upright or hanging.
3. While she spends less time pulling to the left over time, her time spent pulling to the right does not change.
4. Her level of movement increases by a factor of about ten - indicating that she is clearly uncomfortable in even position.

Conclusion

This analysis supports the hypothesis that wrist distance (arm hang angle) strongly predicts the degree to which arching is preferred, and discourages hanging.