markdown: Fix bullet list parsing

Jupyter markdown allows bullet list to start right after the previous element,
without any blank line in between. The default settings of pandoc is to not
recognize that as a bullet list.

Therefore, enable the pandoc markdown extension that makes the blank line
optional.

src/nbsphinx/__init__.py

@@ -1043,7 +1043,9 @@ def filter_func(text):
     input_format += '-implicit_figures'
     v = nbconvert.utils.pandoc.get_pandoc_version()
     if nbconvert.utils.version.check_version(v, '1.13'):
-        input_format += '-native_divs+raw_html'
+        input_format += '-native_divs'
+        input_format += '+raw_html'
+        input_format += '+lists_without_preceding_blankline'
     if nbconvert.utils.version.check_version(v, '2.0'):
         input_format += '-smart'  # Smart quotes etc. are handled by Sphinx