Text this: MM-VSM: Multi-Modal Vehicle Semantic Mesh and Trajectory Reconstruction for Image-Based Cooperative Perception