CLIP-UnmixRS: Vision–Language Assisted Hyperspectral Unmixing With Semantic Endmember Prior Learning. JAAIR 2026, 5 (1).