Why do you say Markdown should be stored in a separate field and not in 'content'? I don't really understand what your issue is.
A few reasons HTML-to-Markdown is messy:
1. The MD you get back can be different to what you put in. If you write a list with dashes, when it's converted back it may use asterisks instead. Same goes for headings, bold, italics, links and more.
2. There was no official spec for Markdown, so different converters may do different things. (This may be fixed in time with commonmark.org but it's a long way off being implemented everywhere.)
3. There is no specification for the formatting for the HTML itself. MD conversion usually uses regular expressions but for the reverse I'm not sure this is possible (at least it's incredibly difficult) so they need to use full HTML parsers which may be slower.
I see no reason to change how it works right now. Like I said above, you should treat the MD stored in the same way you treat plain text. Plain text is not stored in HTML, neither should Markdown.