I can see that it's complicated. Another aspect is that larger diameter tires at the same pressure have lower rolling resistance and that probably explains the BMW choice on the i3. The engieneering choice is higher diameter tires, thinner width versus lower diameter tires, thicker width, as those give the same area contact patch and hence weight at equal pressure. The diameter effect seems to be opposite of the width effect and it's probably something that requires specific experimental investigation.Sure, deforming a thin flexible sidewall is less of a loss than the deforming of a thick stiff tread, but both a wide and a thin tire deform the tread and a thin tire would be deforming less total tread width but at a higher angle which is worse because the tire is losing more of it's shape.
Drag coefficient is still the biggest thing effected by tire width on a car overall and the main driving factor for using a thinner tire vs a wider tire and is why efficiency oriented cars use thinner tires.
When you say drag, do you mean aerodynamic drag even if the tire is covered in a tire well as it is for all normal cars (not F1)? The air being moved on the surface of the treads even in the wheel well?