MoDem: Accelerating Visual Model-Based Reinforcement Learning with Demonstrations