1987 GM Sunraycer solar car, Cd 0.089
 

https://calteches.library.caltech.ed.../MacCready.pdfIf this links, it will get you to an article about the AeroVironment/ GM Sunraycer development.
On page-5, you'll see an incomplete image of the 1/4-scale wind tunnel model they used at the GALCIT tunnel at Cal Tech University.
As of 1987, this model registered the lowest drag of any land vehicle ever measured, @ Cd 0.089.
In the photograph, you can just make out the full wheel fairing at the front wheel, part of a complete package.
Concerns over crosswind gust stability of the 573- pound ( 260.4-kg ) total weight car led to removal of the entire wheel fairing package, which degraded the drag to Cd 0.125 @ zero-yaw.
The car was later yaw-tested for SAE crosswind-averaged drag, with Cd 0.147.
--------------------------------------------------------------------------------------
The wheel fairings reduced drag by 36-counts ( 28.8% ), an approximate 80% drag reduction for the exposed wheels.
In Goro Tamai's book, 'The Leading Edge,' he suggests that full, swept fairings could account for up to a 70% drag reduction of the wheels.
Sunraycers' wheels are more exposed than many solar cars that would follow after 1987.
-----------------------------------------------------------------------------------
Dr. Paul MacCready, who ran the program, estimated that a modestly-powered, internal combustion variant of Sunraycer would return 400- miles per gallon, highway.
At 90- km/h, Sunraycer's total road-load power requirement was 2,000-Watts ( 35.76- Wh/Mile ), at Cd 0.125, 14.222 square feet frontal area ( 1.321 meters-squared ), and Coefficient of rolling-resistance of Cfrr- 0.00653.
-------------------------------------------------------------------------------------

Interesting strategy.

Makes me wonder if the engine out of some small Honda motorcycle could qualify as such "modestly-powered".

Modestly powered in this case probably means perhaps a limit of less than 5 hp.

I guess around 7hp would be enough to provide a comfortable cruising speed for real-world driving conditions.

small MC engine
 

The water-cooled engines would have a BSFC advantage. The HONDA Shadow, 700cc, V-twin, @ 57-hp, 6-speed, would make a Sunraycer-ICE variant like a Le Mans racer.
I started down that road, got about $1,000 in parts collected, then after much meditation, decided that it would be a death trap. Where I live, anyway.
There just doesn't seem to be any right to life in North Texas traffic. And no law enforcement.
The car should have at least the performance of a 250 cc motorcycle. Ability to accelerate out of a dangerous situation, when braking won't save you.

Sure, and maybe could allow minimum impact to the aerodynamics.

minimum impact
 

Yes, that's my conclusion.
The radiator and ductwork could be 'off-shored' anywhere inside the body, unlike an air-cooled engine.
If one was going to pull off a 2%- drag cooling system, liquid-cooling would seem the better choice.
The drag reduction itself, would reduce the heat flux required to be rejected from the engine/ cooling system, a positive feedback.
And the BSFC issue, another incentive.

I guess you're consider NACA ducting to provide air to both exchange heat with the radiator and to the intake charge. I was considering the fitment of either one or more radiators to the outside, while keeping the engine better enclosed, just like some Russian aircraft which had a water-cooled engine instead of an air-cooled and were supposed to be less prone to thermal shock damage while descending too fast.