Concerto: Joint 2D-3D Self-Supervised Learning Emerges Spatial Representations (NeurIPS 2025) | AIDAS Lab | Podwise