Welcome to the Q&A Forum

I'm assuming you can't 301-redirect (and that you still need the sub-directory versions to be reachable by humans)? I'm not sure the cross-domain canonical will work completely. I don't have a good example of a sub-folder to root domain canonical implementation. If the "sites" are identical, it should be ok. Robots.txt is going to depend a bit on how people access those. If there are links to the sub-directory versions, then blocking will cut off that link-juice (and the canonical or a 301 will be better). Blocking the sub-directory shouldn't automatically block the domain it aliases, too, unless for some reason that sub-directory is the only crawl path Google has to the outside domain. As long as they're crawling the outside domain separately, I think you'll be ok. I'm just not sure if Robots.txt is necessary here. Sorry, the devil may be in the details on this one. Happy to take a closer look in Private Q&A, if you want to give out some specifics.