Should input rainfall dictate model timestep? - A ‘commensurability’ perspective