From shared gaze to grounding: Linking joint visual attention to multimodal grounding signals in clinical teaching