metadata: Do not raise an exception if a PDF file has empty metadata

Check explicitly for `info is None`, since an empty dict also evaluates
to false. Otherwise it raises an exception when there is no metadata,
and prevents from generating the cover.

This has started to happen recently with some arxiv PDF's, like this
one: https://arxiv.org/pdf/1605.02391v2

src/calibre/ebooks/metadata/pdf.py

@@ -95,7 +95,7 @@ def get_metadata(stream, cover=True):
             raw = f.read().strip()
             if raw:
                 prints(raw)
-        if not info:
+        if info is None:
             raise ValueError('Could not read info dict from PDF')
         covpath = os.path.join(pdfpath, 'cover.jpg')
         cdata = None