Facial Dynamics in Video: Instruction Tuning for Improved Facial Expression Perception and Contextual Awareness