There is of course a way around the problems cited in the article.

If the referees ran the software on the preprint it would find the same problem.

I agree this isn't as good, but it would be a step forward.