Text this: MoCoUTRL: a momentum contrastive framework for unsupervised text representation learning