Move the unknown token replacement step to download

.gitattributes

dan/datasets/download/__init__.py

dan/datasets/download/images.py

dan/datasets/download/utils.py

dan/datasets/extract/__init__.py

dan/datasets/extract/arkindex.py

dan/datasets/extract/exceptions.py

dan/datasets/extract/utils.py

docs/usage/datasets/download.md

docs/usage/datasets/extract.md

tests/__init__.py

tests/data/extraction/split.json

tests/test_download.py

tests/test_extract.py