Yes, we ultimately felt it was a fool's errand, especially because we're a startup with limited resources so we need to focus on the right problems. When we launched version 1.0 long ago it seemed like this wasn't necessary for a "minimum viable product" considering we're targeting graphic designers who might not have as much coding experience anyways.
There hasn't been much that's swayed us otherwise -- I think it was the right call.
I see that Google's tool also seems to produce rather inscrutable output (see my edit to my original post). This does sound like one of those really hard problems where is pretty much impossible to balance the needs of the machine-modifiability with human-readability.
Thanks again for the reply!