FWICT all warnings are about too many positional arguments, which is unrelated to this change.

@jpakkane, we fixed the "too many positional arguments" thing in the master branch

All green after rebase.

I'd still like to see the protocol used for typing the arguments instead of Any. I've also thought about this a bit more, and I think it would be good to use at least two different return codes, one for reproducibility errors, and one for generic failures that don't mean that there's a reproducibility error. Something like:
0 - everything is fine
1 - there is a reproducibility error
2 - there was an error, but it was not a reproducibility one (exceptions thrown, subprocess failed to run, etc)