Gradient-based Regularization for Action Smoothness in Robotic Control with Reinforcement Learning