Steve,
There are some very good articles on cooling drag on the CAFE Foundation web site. One thing they mention is that the most efficient exit is a 'bluff body'. I believe the info is in 'Local Flow 2' on the site:
There was some work done here in the USA many years ago that showed inlet size can be relatively non-critical to drag, as long as the shape of the inlet lip is done correctly. The technique was to regulate total flow through the system with an exit flap. The concept was to use 'external diffusion', where the air was slowed at the opening, instead of inside the cooling duct. Testing showed that there was little difference in total drag, but duct construction was simplified since shape/surface finish were not as critical (air was already moving slowly when it entered the duct). I think Tracy played with the technique a bit on the RV-8; perhaps he can add some info.
Charlie