You could, but why? I wrote a small package to automate my backups, and transform the content into something readable for myself. It's far from feature complete, even for my simple personal use.
But if you're doing a complete read-only copy of roam, why not go all the way and make it editable, and compete with them instead?
You could put in a lot of work, just for read-only feature parity.
Or you could work some more, and have a competing product.
You seem adamant, that the former is a better path, but I don't quite see why would you do all that work for naught.
(Roam already broke my backup script 2 times, and there are better uses of my time than interfacing with a proprietary product, if there are better alternatives.)
That's as useful as exporting your excel sheets into CSV or exporting your Google takeout backup. Yes you have your data, but can you do anything useful and continue working with it without a big deal of effort? I'd also like to continue working in the same learned workflow/method after I take my data.
I don't particularly want to spend ages trying to find "perfect" when roam is "good" enough. The chances of roam disappearing over night are pretty slim and I take weekly backups of my data from them. Worst comes to the worst I'll dump it into a directory and grep for what I'm looking for.
One more point: the exported data in markdown (and maybe in json too) is lossy. It's not just that the you can't use it out of the box, but that significant parts of it are missing as well.