Provide a counterexample when we cannot calculate a max. expected reward because Pmin < 1

Edited by Arnd Hartmanns