R. Martinho Fernandes:
> Maybe you know a use case for this that isn't the implementation nor transcoding?

What about [de]serialization? I just really don't want to complicate
code_point_sequence with endianness. code_point_sequence should be able
to use std::basic_string as buffer and std::basic_string supports only
native endianness.

code_unit_sequence is for working with encoding schemes. All other code
should only work with encoding forms and be oblivious to endianness.