In identify(), comparaison was done against `str` instead of `bytecode`, due to the lack of `b` prefix ImageInfoTest: add missing test PDF file identification
