I think you're right - I can't see the PLA being strong enough to take the stress. One possibility might be to have a hole drilled in the shaft and a bolt through the hole so the torque is transferred to the outer skin via the bolt. As an aside, square nuts are best for small captive nuts in PLA - much more resistant to spinning in the slot. M3 are 5.5mm A/F and I've got some in stock, M4 and above are ok in hex.

I've put some proper bearings in and the gears are working much better. There is one tooth on each gear that is slightly malformed and that causes sticking but a couple of minutes sanding eased it considerably. Progress is a bit slow at the moment because I'm trying to get the printer working again after adding some hardware to allow me to print without it being connected to a PC (it keeps dropping the connection which is a pain) - all seems ok apart from it not heating the bed which, unfortunately, is a show stopper. If I can't fix it I'll revert to the old firmware but it's probably something fairly simple.

Your new ring gear design is much stiffer and I don't think skipping will be a problem. Engagement seems fine as long as it's within certain limits but when the engagement zone of the teeth moves to the more angled zone near the tip, the forces increase rapidly causing the skipping so a moderate increase in stiffness (and improved tolerances from the new slicer) has made a big difference.