Indeed, the EKF is a mathe-magical thing! But the wierdness that I experienced probably wasn't caused by it.
I've never bothered to dig deeper... but I think one major issue is that the plane scales its control-surface movements with respect to airspeed. Normally this makes sense... as higher airspeed makes a given aileron deflection much more impactful... but in the case of an incorrectly-high airspeed, it kills the control-surface effectiveness.
I still have no idea why it seemed to be asymmetric for me, that day. But the plane was a flying-wing with elevons, no rudder, and small asymmetries in the sensor mounting. So I am not surprised when odd flight behavior happens on this UAS.