Text this: Infrared and Visible Image Fusion Using a State-Space Adversarial Model with Cross-Modal Dependency Learning