Divert More Attention to Vision-Language Object Tracking