Enhancing Trust in Large Language Models via Uncertainty-Calibrated Fine-Tuning