Discussion of the paper by Cule, Samworth and Stewart: \Maximum likelihood estimation of a multidimensional log-concave density"