The core stack is very short on a car alternator, skewing works better with a much longer core so that you get a slot pitch skew with a lot smaller angle. I think you have confirmed a lot of my findings, some skew helps a lot but the skew for very low cog is extremely critical. I had a core pack relatively much longer than yours with the lams not welded, it was in a close fitting tube and I could skew it by putting a steel bar down one slot and twisting it. The thing was very critical and by the time I clamped it and welded it it was not as good as i had originally got it.
I suspect that the job would be far less critical with arc shaped magnets that present a cylindrical surface to the core. SWWP used arc magnets but instead of skewing they staggered the spacing, again I think it will be experimental, the basic maths won't get you close enough, there is too much flux fringing to be able to calculate things exactly.
I don't think crowning the teeth is practical in this case, it does help a bit with salient pole machines such as the F & P but the improvement is not great.
See if you can run it and see what you can get out of it, if you can get way more power than you need then you can widen the air gap and wind with more turns. For now just load it with the voltage that suits it and check the Watts, you can change the voltage later if it has plenty of power in reserve.
Weaker magnets will certainly reduce the cog but will reduce the output just as you will do with the present magnets and a large air gap, with only 8% loss of volts the gap flux is still high.
If you have the original claw rotor you will see that the pole tips are trapezoidal, this is a form of skew and the tips are thick near the disc and thin at the edge, this also makes the effective flux distribution tapered. This is not entirely done for cogging but I have noticed that the cogging on a fully excited car alternator is better than I would have expected.
Flux