Interesting websites, and worth reading. They are for off roaders, with their extra suspension travel, so some of what counts there doesn't really translate for our uses.
Our axles have a shoulder where the splines end, and a thick Bellville type washer goes between the shoulder and the CV. These sites are talking about 930 (or 934/5) CVs and, I suppose, axles. They are discussing the 10mm bolts. They don't show shoulders on the axles. This isn't a bad thing - for off road use you want a longer splined axle and no shoulder, so the axle can float and better deal with the large angle changes. Works fine for street/track use also, but not necessary.
I see one site says use red Loctite on a few bolts, whereas the other says you'll have a heck of a time removing red Loctited bolts, so use blue. Perhaps off road use is tougher in this regard, but for street/track I have not had issues with no Loctite, and I don't go overboard keeping grease off the threads. Proper torque, plus retorquing after some use, works.
They do stress the groove for the outside piece, and the raised end for the inside or star piece. Other than for a reference if you want, I see nothing which makes the way you install the outside piece matter.
I looked through my collection of used CVs (six or so) and used CV parts ( 3 cages, two center stars, and 8 or so ball bearings). Only one of the star pieces had the raised edge. All, however, had about an eighth of an inch of the internal splines milled off to about half the spline depth. This lined up with the one raised end center race. All the rest had the milling, but no raised edge. And the raised edge is at the same height as the non-raised edged ones.
My take: it is the internal milling which counts, although just a little bit. It must be there to make getting the CV started on the axle splines easier. GKN must have gotten tired of the extra machining needed to cut away some of the center race to create the raised center, so they stopped doing that. So don't be surprised if you don't find a raised end on your replacement CVs.
If a guy cares, he will look for the relieved inner splines, and put that toward the axle. As for me, if I can get the CV to go on the axle, I'm not going to care which way is which unless I have spreading out the wear in mind. Though I think I have usually put them on with this relieved part in, even if I don't bother about the groove in the outer race.
Lastly, there is the question about the thick and the thin side of the cage. I measured, and while I didn't use a nice height measurer on a granite stand, and accounting for the fact that one side of what you are measuring is beveled/sloped, it looks like the difference is between 15 and 20 thousandths. Not much. From one of the websites Winders listed I see that the reason to put thin side in is that this will prevent the protruding ends of the splines of the axle from contacting the outer race at extreme suspension angles, and thus damaging the splines and making R&R harder in the future. This I hadn't known. Knowing it, I am not sure I will measure in the future, though. We don't get our rear suspensions to full droop on the street or track absent some kind of accident.
But working with the orientation of these cages might be a way to move the wear spot on the races a little, which is useful when rebuilding.
These cages are no longer of the quality they once were. With a new cage you should be able to snap the balls in, and they would stay put, at least against gravity, when rebuilding. The last couple of years I have found that brand new CVs have loose cages, and the balls don't snap in. How important this is I don't know, but don't fret overly much because there isn't anything you can do about it short of expensive custom work or sourcing or the like.
I did pick up one other thing from these websites: clocking the CVs. The idea is to install one, and then while installing the second to have a wide groove end on one side (the things you are careful to observe when putting the CV back together, so wide inner faces narrow outer, etc) line up with its counterpart narrow groove end on the other side. It appears some think this makes for a better relationship of the ends when driving at large axle angles. Not sure how that would work, though like most of this stuff it wouldn't be hard to do.