Auditing Large Language Models for Enhanced Text-Based Stereotype Detection and Probing-Based Bias Evaluation