As for magnification, if you're shooting all three negatives through the same lens simultaneously, then this cannot be a problem unless you have some other lensing elements in one of the paths. After all, if you've got the image focused on three GGs at the same place, then all three signal paths must necessarily be the same, hence the magnification must be the same.
The biggest problem with using beam splitters to get three color separation in my mind is loss of light in the process. Each pass will give you a one stop decrease, and then the filter factors have to come into play. Even if you start at f/11, you're going to wind up doing no better than f/32 by the time you get to the film plane. And maybe even worse if you've got to introduce a bunch of neutral density in some filters to get them all to match. (OK, Ok, to the purist it isn't really smaller diameter f-stops, but signal attenuation along the transmission pathway. But when calculating exposure I, and probably most of us, think in f-stops.)
The Phillips prism is the Cadillac method to do the separation, but so far I haven't found any unused Phillips prisms laying around in my junk stash. If anybody has a spare one let me know!!
(I even thought of cutting up a bunch of acrylic sheets and making a Phillips prism out of water chambers, but I don't know enough about the physics of light to figure out the dimensions based on the differences between glass and acrylic/water. I know it's different, but I don't know how to correct for it.)
Hence my searches for the two color method. There's far less attenuation in the signal path with only a single split. So what if you do two-color and it isn't true color; sometimes you gotta compromise. It is just a hobby, after all.
And I did track down the filter colors at one point, but I've lost that info.