I can verify that the xslt file is:
b) has an encondig UTF-8
ATTENTION for the the b.
You have to verify the encondig of the file and not the declaration.
According to Notepad++, it is UTF-8 no BOM.
The file is not only well-formed, but it validates against a schema. But it looks like some Java XML libraries have trouble with it, which is why I have to use Saxon instead of Xalan.