LightNav-0: Eliciting VLM Spatial Intelligence for Generalist Embodied Navigation | Light Origins Team | ResearchPod