At the beginning the rendering was invalid because it parsed the entire
content as a single line. For some reason, when I read the ticket
description from the DB, all newlines are returned as CRLF. I don't know
why yet or whether it can or should be changed, but as a quick fix, I
made the handler function filter out the CRs from the text. Then the
rendering is correct.
This matches the documentation of Pandoc, which mentions the readers
assume newlines are encoded as LF.