Sliding to the global optimum: How to benefit from non-global optima in multimodal multi-objective optimization