My thinking process was (assuming open-air construction):
If resistive element of some length is made of totally uniform resistive material, then I think it would always arc between it's ends. That would be the voltage limit (maybe I'm wrong on this).
In contrast - if resistive element is, for example, wirewound, then it would arc between adjacent windings. So sum of all gaps between windings becomes limit.
I was just interested where in this picture falls carbon track pots or carbon composition resistors.
Why someone would want to operate components way outside their specifications is a mystery to me. If you need HV components, then buy such parts. It may be expensive,but that's the way it is.
I'm far from advocate of using parts outside specification, I always pursue a quality design. Sometimes, aside from just doing things, I'm curious about various physical aspects (scientifically). I thought that this one is interesting.