Yes, you could theoretically save 5% of the energy. A small percentage of that will show up as dynamic compression, (and a resulting retarding of the spark and/or richening the fuel curve) but most of it will go out the tailpipe. The other issue is that if you have 33% of the heat going through the cylinder walls, and you reduce the surface area by 15%, a bit will be conserved in combustion, but the cylinder walls will now absorb more of that heat. Keeping 5% more heat in the combustion from your hypothesis is only plausible if the iron and surrounding coolant don't absorb any more of the heat they're being given.
I think the small differences it would make don't offset other factors in what consumers want, like smoothness, quiet operation, small external size, and a flat torque curve. The small percentage of efficiency you could gain with a 15% drop in surface area would mean the public might have to go back to 3 cylinder engines
As a physics and engineering example, you might be able to duplicate this by experimenting with equal displacement V8s and V6s, like a chevy. There exist both 4.3L V6s and 4.3L V8s with identical architecture. The V6 is just the V8 with two cylinders lopped off. The V8 would be hampered by a lot more surface area.
Another nice side effect to your proposal is that the smaller surface means less swept area and therefore less friction.