Text this: A Transformer-Based Multimodal Fusion Network for Emotion Recognition Using EEG and Facial Expressions in Hearing-Impaired Subjects