Another possibility, academically speaking...
It is possible that the fairly slow heading change to the right is due more to a very gradual yaw to the right rather than behavior typical to an un-stalled coordinated turn with its bank angle and associated load factor and resultant turn rate. It is possible at very high AOA for a left roll input to create this situation.
The roughly 2 degrees per second rate, while not imperceptible, would probably not have a noticeable “feel” to it and would probably have had to have been observed on a heading reference scale, especially if masked by a post-stall wing-rock. It’s also possible that the yaw, though favoring the direction opposite the roll input, could be inconsistent in rate, i.e. “stops & starts” in the descending geometric plane.
If the bank was oscillatory in the stall, the 2 or so degrees/sec could have been an average over the roughly 135+ seconds then. Any period of unloading would have stopped this temporarily also. (Roll direction consistent with the roll input is also a good indication that you are actually unloaded.)
In some swept-wing aircraft with ailerons only, this “aileron drag” can occur at AOA’s even prior to the stall and results in some magnitude of yaw opposite to the direction of applied aileron, slowing the roll rate, or stopping it altogether, or in extreme cases causing opposite roll. This is not a spin, just progressive yaw. Manual application of rudder, with neutral or minimal aileron, is used at very high AOA’s in these kinds of aircraft.
ARI’s & spoilers augmenting ailerons prevent this to any noticeable extent pre-stall in large jets. FBW aircraft in a full up FBW mode are far less prone to this anyway, but post-stall with essentially direct control of the aileron surfaces can produce this in the big jets to varying degrees also.