Add support to total batch size argument for transformers transfer learning

[anmarques]

Created the NMTrainingArguments that inherits from HF's TrainingArguments. This class allows one to add arguments to the training script and handling potential conflicts with other arguments. In particular, added "total_train_batch_size" and "total_eval_batch_size" and made sure these arguments cannot be used alongside "per_device_train_batch_size" and "per_device_eval_batch_size."

[spacemanidol]

@alexm-nm given how many commits have happened since you set up this PR I would suggest either starting from scratch on new branch or resolving all the conflicts

[anmarques]

This PR is ready but needs to be merged by hand. Will create a new PR soon and delete this one when done.

[anmarques]

Too much time has passed. Will re-implement if needed