Optimize unicode_subtype_new(): don't encode to wchar_t and decode from wchar_t

Rewrite unicode_subtype_new(): allocate directly the right type.
2 files changed