I noticed that the MarkDown fiter doesn't always parse out HTML
blocks correctly. For example, in this test each tag is sent as
content individually - rather than as a full block.
Instead of
"<ul><li>item1</li><li>item2</li></ul>"
Several calls are made to the subfilter (html) like this:
"<ul>"
"<li>"
"item1"
etc..
This makes it impossible for state information to be processed
correctly. The tests were probably working before because of a bug
in the HTML filter (now fixed).
@Test
public void testUlInTable() throws Exception { // Okapi issue #685
testRoundTrip("ul-in-table.md");
}