Proper Unicode case transformations of single code point

[dimztimz]

Most of the case transformations are done properly according to Unicode specs, e.g. a whole string is made in uppercase.

At few places I use u_toupper() and u_tolower(). Instead I should use something like to_lower_char_at() which is able to do 1 to many case transformations. (eg. sharp s to double S). Make it efficient by minimizing branches and byte copying.