MetaCAT BERT update to tutorials 4.1 and 4.2

[shubham-s-agarwal]

Updates to Tutorials 4.1 and 4.2: Integration of BERT Implementation

• Tutorial 4.1: No changes were necessary since BERT utilizes its own pre-trained tokenizer. A note has been added advising users to proceed directly to Tutorial 4.2 if using BERT.

• Tutorial 4.2: BERT implementation has been added, showing training on the given dataset. Unlike BiLSTM, which fine-tunes a previously trained model pack, BERT is trained from scratch.