rewrite gap buffer to use bytearrays instead of array.array('u', ...)

[kcr]

because arrary.array('u', n) is deprecated and bytearrays are part of the core language now

[asedeno]

I've looked into this and it's frustrating because array.array('u', n) is a container of characters (PY_UNICODE, and later wchar_t), not bytes, so to switch one would need to manage characters at byte level with some encoding. It may be easier to detect an array.array type that is 4 bytes and use that along with UCS-4/UTF-32. I'm hopeful that https://bugs.python.org/issue36299 will yield an alternative before this becomes an issue.

[asedeno]

Python 3.12 will have a UCS-4 array type of 'w', so there's a reasonable transition path: if we find 'w' in array.typecodes then we use it, otherwise we keep using 'u'.