In some cases, it would be more optimised to first do the boolean operation then rotate the solid.
For instance in the casing for PrincetonD TF coil, each component is a union and a cut of 3 components. This makes (3 + 1) x nb_of_casings boolean operations.
By first union/cut then rotate, you'd only union/cut once then rotate. Which would make 3 + nb_of_casings operations and would save up much computational time.
A nice feature would be to control what we do first