Few-shot Adaptation of Multi-modal Foundation Models: A Survey