use file extension for determining mimetypes on default

should be more reliable for files that "look like" other types based on their magic numbers such as odf and docx files
also a lot faster then the old way