Text this: Degrade or Super-Resolve to Recognize? Bridging the Domain Gap for Cross-Resolution Face Recognition