LACA: Improving Cross-lingual Aspect-Based Sentiment Analysis with LLM Data Augmentation