Text this: Adversarial Robustness of Self-Supervised Learning Features