LLaDA-VLA: Vision Language Diffusion Action Models