Chris,
The Thorp T18 uses the cowl side exit, for that reason. The counter argument in some aero documents is that it messes with airflow over the top of the wing, where you want flow as perfect as possible.
One if the reasons given in some docs about the choice in the P51 was that it moved the inlet far enough aft to eliminate the prop pulses from entering the duct. The reason for *that* is that the pulses disturb flow in the duct; some of the energy hits the core and reverses, trying to push air back out of the duct. Note that circumstances are different in a cooler duct vs an induction inlet. When LoPresti did his 'Holy Cowl' work, the induction inlet almost touched the trailing edge of the prop blades, BUT, the prop was clocked so that each pulse raised pressure at the intake valves just as one was opening. At least that's what his advertising claimed.
Submerging ducts inside structure requires structural analysis; relatively easy on tube/rag because the fabric isn't structural, but can be hazardous on semi-monocoque fuselages like RVs, Thorps, etc if load paths aren't analyzed carefully. Probably easier on fiberglass, but still deserves attention.