As to rewinding a motor for a higher voltage, the controlling effect is called "stacking factor", the number of turns you can fit in a space compared to the theoretical number if every turn could be perfectly even and in line with other turns. In some cases such as transformers, the advantage is to the lower voltage winding because you can use rectangular wire and make nearly perfect layers. In motor armatures you are pretty much limited to scramble, AKA random winding. In perfect layer windings, any wire size will fill the same space with the same amount of copper, the only differences will be secondary ones like the ratio of the insulation to copper area.
In the use discussed here, you could run the motor at a much higher voltage by pinning it to the middle of the voltage, 250-0-250, each side never exceeding 250 volts to ground, the controlling factor being the insulation between the coils and the armature laminations and the turn to turn insulation. You could run near twice the normal base speed and only field weaken on a smaller portion of the range. You would need current control on the armature supply to avoid slamming it with 500 volts in a start. That is one reason I like saturable reactors which are intrinsically current limiting. The common form, not rapid reset reactors like the Hendy lathes use, can be dead shorted and will only supply the programmed current, giving a softer start than the original systems, less sparking on the brushes, etc.
In my experiments I found that regulation of the field supply was critical. At the top of the speed range speed control is flaky with a small variation of load or field voltage causing speed fluctuations.
Bill