I forgot about this, but the 2014 study itself provides some help on this. Despite the methodological problems as far as the measurements of the number of non-citizen voters, it's probably still a reasonable source for partisan affiliation of non-citizens.
They found that 81.8% of non-citizens who claimed to have voted in 2008 reported voting for Obama. It's on p. 5 of the article, which I've posted
here.