I don't think there is an answer to your question. Many different variations in spacing and coil size work perfectly well. Somewhere there must be an optimum but I doubt that anyone has done a computer analysis sufficiently accurately to reach the perfect arrangement.
The most efficient winding may turn out to be a multilayer winding with overlapped coils and this would need very different spacings. It is also more difficult to wind and requires a lot of skill to do it. In most cases the end result will be no better and in the hands of less than a very skilled winder it will be worse.
I could post pictures of waveforms if I could be bothered to transfer data from an a old computer but it would be rather pointless.
I can show you cases where the 3 phase waveform is virtually the ideal sine waves displaced 120 deg and I can show you versions where the 3rd and 5th harmonic are as large as the fundamental, some in phase and some out of phase. the results are very flat topped waveforms and very peaky ones. When this is rectified and used to charge a battery it is of little consequence as the rectifier murders the waveform anyway.
None of these alternators produce anything remotely resembling sine waves into the ac side of the rectifier. For star connection it makes no real difference but it does play havoc with delta connection and that is best avoided in any circuit with non linear loads.
Don't get too hung up on voltage waveforms, altering magnet and coil spacings will have an enormous effect on the voltage waveform but has very little effect on the power into a battery.
Flux
