I've been using pandoc + rst + beamer for producing very large (1000's) training slide decks, and pdf2latex is so slow that using a CI to generate the 10's of decks configurations in parallel makes for a very valid use case, so I implemented just that... I don't know what I'm getting at with this little story of mine, but if you actually scratch that urge at some point I'd be interested in seeing the result, especially regarding distribution, which I didn't set out to implement.