Performative Reinforcement Learning with Linear Markov Decision Process