DiffStitch: Boosting Offline Reinforcement Learning with Diffusion-based Trajectory Stitching