Which is why I was so puzzled as to why the file(s) would take on the requested "MimeType" after changing that line.
From my experiments (and the documentation), file can only guess at the encoding by the presence or absence of specific codepoints. If your file has only those UTF-8 codepoints which also fit the Latin-1 encoding, perhaps it's all too happy to report the file as Latin-1. In my experience, iconv doesn't add a UTF-8 BOM to the start of the file. When I did that with Vim, file was a lot more specific.