No this concept of fighting currents and wasted power is not right.
If the geometry is not ideal, all you do is generate less volts. For the same load you therefore produce less power. It is not a loss in the sense that you mean, you don't produce it and therefore it doesn't come from increased input power as a loss would.
In any machine without slots you will only get the perfect theoretical result with a coil with a single wire, once you start to produce a normal coil with parallel wires ( finite leg width) you can't link all the flux instantaneously. What you end up with is a distributed winding where you progressively link the flux. The peak of the emf is lower but the turns are still contributing at different times.
Far better to think of flux linkage than worry about legs and things fighting ( if you must worry about that it is the voltage that fights not current).
If the hole in the coil is bigger than the magnet you will link all the flux at some point in the cycle but not instantaneously unless the coil has sides only one wire wide. If the hole is smaller than the magnet then not all turns see the full flux linkage but you still gain some volts. What matters to the output you get is volts per unit resistance. If you do something that increases volts without increasing resistance you gain. With these turns inside the hole dimension of the coil you don't gain your full amount of volts but these are very short turns and add little in the way of resistance so you come out better as long as you don't go crazy. Turns of negligible area will link little flux so there comes a time when you start loosing.
Similarly on a single rotor you are right, the flux doesn't link much of the back part of a thick coil. Within reason using thicker coils gives you more room for thicker wire ( low resistance) but beyond a certain thickness the volts induced in the back turns becomes too low and you are adding resistance and you loose out.
For a single rotor the best may be with coils about one magnet thickness, certainly thicker is likely not to be any improvement, thinner may actually be better. The thing that is critical in a single rotor is the distance from the magnet to the coil face ( don't confuse this with air gap). There is no defined air gap with a single
rotor, the longest flux path is via infinity and the shortest is the near point between points of opposite polarity on the magnet.
If you have losses with your winding it will not be from what you describe, but it can happen if you parallel circuits with any voltage difference. Parallel coils are problematic, delta circuits will definitely produce loss from harmonic circulation.
You say you have delta and yes it will suffer circulating losses and more so once you rectify and increase harmonic content. Don't forget that you will have winding resistance and this will drop your efficiency even without circulating losses. If your machine is something based on Hugh's 4 ft machine I would expect your efficiency from resistance loss to be down round 50% at something like 200W out. Delta will make it far worse. Jerry connection should prevent circulating currents below cut in but they come back when the rectifiers conduct and show as loss just as in delta.
If you have drag or losses with open circuit coils or star connection you have eddy losses from wire that is too thick or shorted turns. I don't have enough details of your winding or enough information to understand why you think you have these losses to comment further.
Flux
