Towards Explainable Evolution Strategies with Large Language Models