Enhancing Continual Learning in Visual Question Answering with Modality-Aware Feature Distillation | Synapse