Enhancing Robustness of Retrieval-Augmented Language Models with In-Context Learning