> My reading of their paper was that they want to encode non-UTF8
> sequences into UTF8 paths in a JSON file. I don't think they should take
> that path, because it loses too much information.

What if non-UTF8 part will be stored in the same way HTML encodings do?
So we would have UTF-8 string as the name of encoding such as "WTF-16"
for NTFS and array of numbers that are "abstract units of text" (code
units for UTF, characters for US-ASCII, not sure about other encodings).