Reward-Free Model-Based Reinforcement Learning with Linear Function Approximation