Text this: Enhancing Sign Language Recognition Performance Through Coverage-Based Dynamic Clip Generation