Did charge the battery completely, and obviously now it works. Test 1 showed a shunt off by 0.54A. Test 2 gave nothing - I don't have a setup allowing for this kind of test. I did spin the wheel in the air, but the load was not high enough to cause significant voltage drop amongs individual P groups. Bypassing the bms works fine. Didn't try changing the limits yet. I will let the battery balance-charge a few more hours and then go for a longer ride to empty it again. Let's see when the cutoff will happen again.