Text this: Efficient Sedimentary Facies Recognition Using Vision Transformer and Weakly Supervised Deep Multi-View Clustering