This function takes an environment and an optional step size, and returns an object of type IloMetaHeuristic which implements a greedy local search improvement mechanism. If the step size is not supplied, it defaults to 1e-4. When applied to a neighborhood exploration goal, the returned object rejects all movements that do not improve the cost variable by at least step.