Text this: Evaluating masked self-supervised learning frameworks for 3D dental model segmentation tasks