V2X-DGPE: Addressing Domain Gaps and Pose Errors for Robust Collaborative 3D Object Detection