Learning domain-agnostic visual representation for computational pathology using medically-irrelevant style transfer augmentation