Do not assume the characters are ASCII

[bieleluk]

Many functions assume that the characters are ASCII only.
I.e.: longest_prefix, split_two_lines, longest_suffix

We should make them stable also for the generic Unicode characters.
As the functions are mostly used for rendering, the better approach seems to be using the TextLayout instead or https://doc.rust-lang.org/std/primitive.str.html#method.split.