Towards Few-Shot Identification of Morality Frames using In-Context Learning