Commit for documentation update for meta annotation

[shubham-s-agarwal]

Documentation update for meta annotation training to include training for bert and 2 phase learning.

[mart-r]

Regarding meta_annotation_training.ipynb:

• Is it just BERT or just LSTM MetaCATs? Would a mixture also work? Or does it simply not make a lot of sense? Maybe document that little bit - i.e the user should run 1 of the 2 code blocks, and not both of them; and potentially how a user could know whether the MetaCATs they've got are BERT or LSTM based.
• Is there a reason the saving of the results was omitted?

Regarding meta_annotation_training_advanced.ipynb:

• It feels more like a tutorial, or perhaps even a showcase. Though there is value in having both a tutorial as well as a script to be used in deployment. I just want to make sure this is how someone would use

[shubham-s-agarwal]

I've added a code block using which the user can see which model is used for that particular model pack.
I've also added the saving results back in.

[mart-r]

Looks good to me