I'm the author of an ordinary software-in-your-computer 1541 emulator — the concept is nothing new, the chips are very well understood. Regardless of the article's comments about the 6522, doing it as a proper real-time process compatible with the original signalling is the interesting bit.
That being declared, I disagree with the reasoning for your guarantee. Emulating proper physical timing is pretty trivial, especially if you've a whole Ghz-level ARM at play.
To my mind the main obstacle is more likely to be the Commodore file formats: the most advanced one, G64, is a "raw bit stream" that permits multiple speed zones per track and atypical track lengths, with half-track positioning but all data within a speed zone is implicitly perfectly clocked and of perfect amplitude. So amongst other things, weak bits and fuzzy bits are not conveyed by G64.
Some of us have a full pretend PLL in our emulations; I'll wager that was omitted.