Suggest to correct __init__.py in logppt/models/__init__.py for bert and gpt2 models

[Tasneem91]

When the required model is bert or gpt2 the functions add_label_token_bert() and add_label_token_gpt2() take 3 positional arguments where 4 are given. So we need to correct the function call and remove the parameter wo_label_words, where it is not needed add_label_token_bert(model, tokenizer, label_map) add_label_token_gpt2(model, tokenizer, label_map)