Update documentation to clarify that UTF-8/16 checking is done on complete strings before any other processing.