Continual Text-to-Video Retrieval with Frame Fusion and Task-Aware Routing | Read Paper on Bytez