Using ball races for the main yaw may help slightly but it does also encourage it to yaw about more. Again I am not sure it is an advantage  as far as control is concerned.

There seems to be no way to make furling a precise process, variations of 2:1 in the current in the region of furling seems to be unavoidable and the thing ultimately has to survive the highest of these currents. The relative heating in the high peaks does depend very much on the site. Turbulent sites may have little energy in these peaks but on a good clean site in a good wind area the high peaks may continue for many seconds or minutes. Usually as the wind becomes more severe most machines drop power beyond furling, these tend to survive but those which hold peak current or worse those where current continues to rise are likely to cook.

One big problem is that variations in construction does seriously affect the prop speed, power curve and the furling characteristics.

The axial machine is very forgiving in that it will work with vast differences in construction whereas a badly wound radial just won't work, the thing will not go together if the coils won't go in the gap.

Other factors such as line resistance also also have similar effects and when several factors come together you may get a machine that performs badly but never burns out at one end of the range and at the other you may get wonderful performance with short life.

If you can't have quality control then the only cure is to drastically over build it, expensive in the short term but with little effect in the long term running cost.

I think under these circumstances with limited facilities and skilled labour adding bearings to tails any yaw would drastically increase the risk of failure rather than help.