I have to agree with JohnR in principal here. It's not typical for a movement (especially a quality movement like that) to encounter a terrible error like that. Escapements like these are designed to be isochronous to the extent that they should not vary considerably in rate with the amount of energy that is transferred into them, unless they are driven into banking. I exercise most days and I've never encountered any of my watches doing anything like that. Maybe I've been lucky.
Vsauce, of YouTube fame owns a backward running quartz. He visited "Adam Savage's, Tested" shop and Adam and him compared watches. Adam has an omega "beater", a problem I wish I had! lol. It's pretty clear the Vsuace is not a watch enthusiast, given the rough handling he gave Adam's watch while he was looking it over. He shows off the backwards watch HERE and the entire video is worthy of a watch HERE as well.
I gave it a go, very carefully and it worked until I went to clip it back on to the watch. As I clipped it on, the bezel cracked again just after the repair and a small piece fell out. It is very brittle. I tried again but it didn't turn out so well. Fortunately, JerseyMo came to the rescue and sent me a nos bezel. Thank you JerseyMo and thank you too Manodeoro for your advice. It is much appreciated!
In dynamic state and with associated random high amplitude, hairspring may stick.
A video of HS as you manually complement barrel power, would reveal such sticking. Should you set up for vid, I try to capture the guard pin as well.