I don't know how much clearance you need on a large machine, much depends on your bearing set up. I manage with about 1mm on smaller machines but the last thing you want on a high voltage alternator is a rub between rotor and stator. 5mm each side is going to cost you a lot in magnet but it may be a cheaper option in the end.
Yes thinner magnets and a thinner stator with more poles will give you more surface area and almost certainly better cooling but again the large clearance gap will knock your flux density worse than with thicker magnets, it's a trade off again. I don't think you have a heat problem now but if you are worried then this will help as long as the thinner stator is mechanically strong enough.
Coil dimensions and magnet spacing is a grey area. If you think in terms of single turn coils then you are forced to the conclusion that the coil sides should be placed at the centre distance of adjacent magnets. If you think of that stupid E=Blv equation it keeps forcing you back to this conclusion but in real life you have to end up with a distributed winding, not all turns can be in the same place.
It is better to think of flux linkage and as long as turns of a coil don't stray into the reverse field under another magnet you will still gain some volts with turns way outside what seems a reasonable area.
Even making the holes smaller than the magnets doesn't prevent those turns linking flux, they just don't link all of it. There is ultimately a trade off between volts generated and added resistance to get it so it becomes near impossible without computer simulation to hit an exact compromise.
I find it better to have gaps between the magnets at least the magnet width any less is wasting valueable copper space. If you go vastly too far then the length of turns in the coils will be adding too much resistance on the outer turns so I compromise with gaps equal to magnet width at the inner circle of the magnets.
You don't loose much volts by squeezing the coils a bit triangular at the inner circumference of the hole and those short turns have little resistance and the gain is made up by being able to get more turns in.
If you must make the coil legs a pole pitch you will have to resort to overlapped coils, it only happens on the outer few turns of a single layer winding, but you will come off much worse with a full pitched 2 layer winding in terms of greatly increased resistance from the end windings. For axials the simple single layer winding works out better and is infinitely easier to wind. The same doesn't apply to radial designs.
Flux