The threaded ends of the axle aren't visible in the photo, because the Primato happens to come with an axle that is just long enough that the axle nuts have full thread engagement when there are spacers for 130mm, but there's nothing poking out. Yes, I have similar setup.

yeah good point there - its definitely more leverage, but not as much as in a freehub design.

Bolts can only take tension. You should be able to replace the axle with a cable, given that it is properly tensioned. Compression forces are supported in the following order: drop-out inner surface, (spacer ring), bearing inner ring, balls, bearing outer race, hub shell, reverse for other side.

Assuming a perfect axle that does not elongate under tension, and sufficiently large friction between spacers and other parts the hub with or without spacers is equally strrong.

But with a real life axle elastic elongation is possible and that would allow spacers and hub shells rotational symmetry axes and normal vectors of drop out surfaces to loose collinearity. That is, the hub shell may twists (eg lifting on the drive side and dropping on the non drive side) whilte the spacers move in the opposite direction.

It would not be bad if the hub were held together by a cable, as the hub would simply collapse.

However, a rigid axle would be subject to a bending momentum.

Each spacer ads degrees of freedom. The momenta depend on aspect ratio of spacers.

The axle cannot be made to take bending momenta, as it has no load bearing interfaces with anything but the bolts on either side. (It cannot transmit forces directly to the frame, as the only interfaces would be the drop out slots with the threaded surface of the axle.)

The whole thing really is a major problem with rigid axles with bolts. A thin, flimsy axle, with a good QR is way better here as it would at least have the decency to fail. With a strong solid axle one might not even notice the issue and runs the terrible risk of a bodge.