This is far from a complete answer, but I want to point out that you need to recognize that the motion of any given part (e.g. molecule) of the aircraft through the airmass creates an apparent wind or "relative wind" in the opposite direction. If an aircraft is changing bank angle (rolling), the "relative wind" actually becomes "twisted" into a corkscrew-like shape. For example, the wingtip that is rising will feel a downward component in the relative wind, while the wingtip that is descending will feel an upward component in the relative wind.
This link should be of some interest -- note that the green arrows labelled "motion" are not both pointing in the same direction, and thus the direction of the local relative wind (which is opposite to the direction of the "motion" arrows) is not the same at each wingtip, and thus the angle-of-attack of each wing cannot be the same. The angle-of-attack of the rising wing has been decreased (or in extreme cases, has actually become negative), and the angle-of-attack of the descending wing has been increased. This creates the "roll damping" phenomenon-- a roll torque acting to slow the roll rate-- which can be loosely described as an inherent aerodynamic "resistance" to rolling. This phenomenon is central to the dynamics that you are trying to model.
From the phrasing of the question, it seems that you already understand much of this, but I just thought that this answer might put it into a slightly different perspective that might be helpful.
PS the link noted above leads to a discussion of yaw dynamics (adverse yaw), but that's not really what I'm trying to focus on here. I'm just using it as a useful illustration of the "twist" in the relative wind due to rolling, and the resulting "roll damping" or "resistance" to rolling, regardless of whether "adverse yaw" is also entering the picture in a significant way or not.