PDFExporter currently has its "output_mimetype" traitlet to be

'text/latex', from its superclass.

This causes issues when using "Export to pdf" option in the notebook,
as the notebook web application uses this traitlet's value as the
mimetype of the attached file.

nbconvert/exporters/pdf.py

@@ -64,6 +64,8 @@ class PDFExporter(LatexExporter):
 
     texinputs = Unicode(help="texinputs dir. A notebook's directory is added")
     writer = Instance("nbconvert.writers.FilesWriter", args=(), kw={'build_directory': '.'})
+
+    output_mimetype = "application/pdf"
 
     _captured_output = List()