I think the indicator in part 2 is a good with a few caveats. Chatter from rotating the ball could be factoring in. I doubt it's much, but it's something to at least consider and be aware of. I don't see any other way to do it that could be done "at home" so to speak. A solid base is important, 3d print might not be solid enough for super precise work, but again, probably good enough for home work. Granite is the de facto surface when inspecting things, in most shops it's precision ground slab from 6" to 3' thick, maybe a piece of countertop would be a solid base without flexing and such as would metal. Just hot glue the bearings onto that, but try to make sure the metal is actually touching the granite so yo don't have flexing happen... or use epoxy. Honestly with a sphere I don't think the 3 bearings being slightly different would have much effect, you're only touching 3 pinpoints on them anyway and essentially checking runout. How much deviation from a perfect sphere is noticeable? I bet if the balls are in manufacturer specs, humans would not notice, especially factoring in cloth, humidity, etc, so maybe a size gauge like posted earlier would be more important, making sure the balls are the same size.
Is it overkill? Probably, but it's a fun project and eliminates some unknowns.
I have access to a cmm at work, but there is a strict policy about not using it on your own projects, even if you're off the clock. Huge money in the machines so I understand it. They make me nervous to use even though it's my everyday job. With the cmm, you could scan the surface with a probe, taking tens of thousands of "points." I think the thing is accurate out to .0005mm or something, just stupidly precise and there are some that go even further. It lives in a climate controlled room. You have to leave the parts in the room for like 10+ hours to get them "quenched" to the room temp so expansion and contraction don't factor in. There are optical scanning heads and such, but if you're trying for the most accuracy, contact probes are still considered the best. I forget the resolution, but non touch optical probes just cant get the accuracy of touch probes. We're talking stupid precision though, well beyond the realm needed for most applications. If you had an extra person in the room, just their body heat can throw things off. It's pretty amazing and most cmm labs are designed with a maximum number of people, heat from electronics and monitors figured in, the works. All of this is software controlled, so if for some reason say the temp went up half a degree, the software can compensate for this kind of thing.
It's above my pay grade to figure this stuff out, but I've heard of testing where they could tell if a different brand of wipes were used in cleaning since they were picking up tiny fragments from the wipes. Some of this stuff just blows my mind. Imagine testing jet turbines and such, they do it, but damn would that be a tedious process! That's one reason why high tech stuff like used on space shuttles is so expensive, they have testing equipment worth millions upon millions of dollars. I'd hate to see what they charge per hour on an inspection that precise!
Anyway, keep up the good work, pretty interesting stuff going on!