Text this: Safely Imitating Predictive Control Policies for Real-Time Human-Aware Manipulator Motion Planning: A Dataset Aggregation Approach