Barycentric Interpolators for Continuous Space and Time Reinforcement Learning