- Use memory view for hashlib algorithms (closer to python 3.11's filehash API in hashlib)
- Remove `sha1_fast` (realized it doesn't even hash the whole file, it just does the first block)
- Add support for custom file filters
- Update docstrings
- Update tests