Right, we cannot forget the 3-dimensionality of the situation. I remember reading in Hucho about the Schlör "Pillbug", that they predicted a Cd of 0.18 theoretically based on the 2D profile, but after they built it they measured 0.13 (they had to tweak things like the height).
My best guess is that the air flow around the sides is how it ended up being better than the 2D profile had predicted. BTW, here's the drawing from the Hucho book of the Schlör "Pillbug". I think the lofting lines are not completely accurate, as I started to make a 3D CAD model by tracing them, and it does not look right.
The other thing that makes me think this is only approximate is the layout of the windshield glass -- the car had a center section of glass, whereas this drawing shows a pillar in the center of the windshield. And I'm fairly sure that the 1275mm height is for the body of the car; not it's height above the ground.