Inhomogeneous target-dose distributions: a dimension more for optimization?