vtkm::Math Pi() operations should provide float versions

currently Pi() is only represented as a double value, and we should provide float versions