The closer the fit of the cowl to the fins (assuming the engine design is finned correctly) the more efficiently you can cool it. But efficiency in a stunt plane isn't super important since extra cooling drag is overcome by our typically huge power margin. For contrast, compare to an F2C model running a diesel engine. The clearance is set around two or three thicknesses of masking tape when you finish the fuselage. And the opening is designed to a minimum so that you never have excess air (excess drag) being force thru the cowling.
Make sure that you get rid of sharp edges in the cowl where you want the air to flow around something. Deadheaded flow is not good. The way to think of this is that aerodynamics are in play inside the cowling as well. Cooling the back side of the cylinder should be a priority, not just hammering the front and leaving dead air in the back.
As far as the outrunner motors, putting half the outer housing in direct airflow as the housing rotates will give you uniform circumferential cooling of the magnets since the heat flow is much, much slower than the rotation rate. (ie. you won't be able to measure a temp rise in the magnet while it is in the "shadow" of the profile fuselage.) Something that I have done on profile race planes is to core a circular duct thru the fuse into the recess where the bottom end of the engine sits in otherwise dead air. Just use a sharpened 1/4" brass tube and put it in at a shallow angle to the fuselage axis. No reason you can't do something like that on electrics, too.
And, since the magnet array is also exposed to the internal airflow, that is a primary cooling mechanism, too. The real heat being generated is in the stator windings. The cooling for that would seem to be almost entirely via airflow in thru the front of the outer can. Again, dead air is a killer, so having a smooth path behind the motor for the air to exit is important.
For motors, they usually are spec'd to survive a certain max winding temperature. It is a situation where time-at-temperature determines when the winding insulation will break down and things short out. The modern "magnet wire" typically uses some kind of epoxy "varnish," which is way, way better than the old kinds of varnish insulation. Still, it breaks down over time. The other heat issue in the motor is the demag due to high temps. The strongest magnets often have a lower Curie point, meaning if you overheat a motor beyond the working temperature of a neodymium magnet, it "recovers" to a much lower energy product rating. Samarium cobalt has a significantly higher temperature rating, but most will have a lower energy product value.
From a theory standpoint, for convective heat transfer the variables are essentially the delta-T between the hot surface of your component and the incoming air, the heat transfer coefficient, the area of the hot component exposed to cooling flow, and the air mass flow rate. Improve any of these and you improve cooling. So having a huge clearance inside the cowling where the air can most easily flow thru the center of this gap--and very little of it thru the fins, think boundary layer here--is really inefficient. We get away with it in many cases because we have inlet and exhaust apertures that are far larger than needed. And if it is working, no one really complains on a model. But aerospace companies spend huge bucks to make their designs efficient in this regard.
Dave