This function returns a goal which randomly permutes the search tree branches of another goal in the environment env. The idea of the perturbation branch selector is to swap, with a given probability probability the branches of each IlcOr constructed by the goal goal. In this way, randomized versions of standard goals can be created, which is a simple way to create randomized populations for evolutionary algorithms. This goal draws random numbers from the generator of the solver on which it is used.