LLM-based Text Simplification and its Effect on User Comprehension and Cognitive Load