Support subword and word language models

.gitattributes

dan/datasets/extract/__init__.py

dan/datasets/extract/arkindex.py

dan/datasets/extract/utils.py

dan/utils.py

docs/get_started/training.md

docs/usage/datasets/extract.md

docs/usage/predict/index.md

docs/usage/train/language_model.md

requirements.txt

tests/data/prediction/language_lexicon.txt

tests/data/prediction/language_model.arpa

tests/data/prediction/language_tokens.txt

tests/test_extract.py