I interpreted the OP's question as "Why not use UTF-8 as the secondary encoding in these types of hybrid string classes which use a narrower representation when

I interpreted the OP's question as "Why not use UTF-8 as the secondary encoding in these types of hybrid string classes which use a narrower representation when possible", not "Why not convert all strings to UTF-8.