Collecting mementos: a multimodal dataset for context-sensitive modeling of affect and memory processing in responses to videos