Unless someone comes up with a documented requirement that metadata is not allowed to be UTF-8, my code changes (PDFString wrapper for PDFStr and PDFUtf) allow UTF-8 metadata for fields. I can understand CreationDate and ModDate having very specific field formats with no place for non-ASCII characters, but for everything else it seems to work fine. The pdfinfo utility mentioned prints out the metadata OK, except that wide characters are missing. Since this is a console text output utility, it's not surprising that UTF-8 characters would not be handled. However, there is no error message.
Until and unless I'm informed that UTF-8 is not permitted in any metadata field, this fix will be in the next release (3.013). If it is necessary to ban UTF-8 in metadata fields, update PDF::Builder::Basic::PDF::Util to remove 'm' code from the UTF-8-allowed list.
Further note that direct use of PDFStr() and PDFUtf() calls are now discouraged — you should use PDFString() with the appropriate usage code.