Calibrating LLMs with Information-Theoretic Evidential Deep Learning