Just to pick a nit and maybe make things more clear (or muddy 
) I think this should be slightly rephrased:
There is always a substantial bike/rider interaction effect.
) I think this should be slightly rephrased: There is always a substantial bike/rider interaction effect.
I am not aware of any evidence to support that claim. (Indeed, since it is hard to assume a cyclist's posture without actually sitting on a bike, I'm not sure exactly how you would go about testing that hypothesis...although the fact that the drag of downhill skiers in a crouched racing position is comparable to that of a cyclist certainly implies that it isn't true.)
Older designs such as the Lotus, Zipp, Softride, etc., were often included in wind tunnel tests, but again, only the Trimble showed clear-cut evidence of an interaction effect.
Edit: Actually I think you probably could attempt a test like that, but it would require a pretty big investment in welding and machining. Simply set up the three points (seat, cranks and armrests) with a space frame that is very far from the rider. Then test the same rider in the same position on the actual bike...and the bike solo. As long as you keep the support structure for the "fake" bike frame far from the rider it would be functionally similar and might be a "good enough" approximation to see if there were significant differences.
