Personally, I think your suggested idea is not going to be that effective, because you have to escape every space-separated word that could possibly be evaluated, that way (as opposed to, for example, using a custom $escape(<whole line>) identifier which could be built with $encode+$decode), so escaping whole arbitrary lines will still be a pain.
I still think that a much more elegant solution would be a non-evaluation switch to the commands that double-evaluate by default (
link)..
