AnoleVLA: Lightweight Vision-Language-Action Model with Deep State Space Models for Mobile Manipulation