Comparing Human Expertise and Large Language Models Embeddings in Content Validity Assessment of Personality Tests