Fix: fix c_char type bug and add source dist

[mkhazin-r7]

There is a bug, where we call the function pyo3::ffi::PyUnicode_FromStringAndSize with the wrong parameter - i8 instead of c_char. This is probably because the documentation also wrongfully used i8. c_char can be i8 or u8, depending on the system, so that's the reason why it compiles on some systems and doesn't work on others.

I also added a source distribution, to allow installing with pip on systems where the precompiled release won't work, like the python3.11-alpine docker image.