I found another HTML content detected as markdown -- I suspect the series of underscores in Summary. I don't think it did this in the last version, but I'm not 100% on this.
Thought: Is it possible to make the presence of HTML tags override the presence of markdown syntax for the detection?