There's really no reason not to do this especially since sub-codepage already defaults to auto. Also change logging in charset_conv since telling us that the data is UTF-8 if the passed codepage value is "auto" or "utf-8" is really not useful information (that's the expectation).