I uploaded a rather large file that was saved using Notepad++. In the info line, the editor claimed the file was UTF-8, Windows (CR LF). Upon uploading, the site said "Failed to read... Are you sure it is in UTF-8 format?"
So I used the editor's menu to convert the encoding to "UTF-8 with BOM", and then tried again, with the same error.
Then I used the editor's menu to convert the encoding to ASCII, and it uploaded successfully!
The file did contain some strange characters. It was the same sequence of three characters in multiple places. The first was a Lambda, which did not appear to be converted to UTF-8 properly upon export from Excel, followed 0xC7 and 0xF4 characters. The last two are valid UTF-8 characters, which are variation C and o, respectively.
However, the file still could not be read after removing the lambda character, leaving the 0xC7F4. I still had to convert it to ASCII in order to upload it.