I think you are misunderstanding the point being made: The _only_ random chance that is part of current models is "quantum uncertainty at big bang time" and we can give upper bounds for the variations that can be explained from that. So what's really being said here is "We found a significantly larger discrepancy between Milkyway-corotating galaxies and Milkyway-counterrotating galaxies than can be explained by ~initialisation randomness"
No it can't. Turbulence does not introduce net-angular momentum. It just (re)distributes it. And the scale on which that "mixing" can happen is limited (essentially the speed of sound is smaller than the expansion of the universe in the early universe). So on large enough scale, it (~any vector value) must be add up to ~zero (up to the initialisation uncertainties). Or one of our fundamental assumptions is wrong. And that's why this is so interesting
I don't have anything more than Newton's third law for you and that its effects also hold in general relativity. The "far away parts of the universe are disconnected" is from my astrophysics courses back in university and the number of lectures to the cosmic microwave background, mostly coming from people discussing the Planck mission.
But.. no, I don't have a convenient citation for you. And at least for the "angular momentum is conserved thing", I'd be surprised if you'd find a google scholar paper, this is early GR