Why don´t you make simple non-contact supports, of high accuracy, zero contact, pwr-4 rigidity and low cost ?
Make some surface-inlets, shapelets, of near -net shape, of appropriate size, ie 50" inside radius as you said, as near as cheap-easy-practical.
Suggest about 50x50 mm squares, or == 60-80 mm D round, turned to near-net shape of the ball surface.
Drill 30-40 holes, about 4x D of the hole, in depth (reservoirs).
So maybe 1-2-3 mm D holes == 4-10 mm deep, in the pads, on the inside, and a bit smaller through hole, say 0.6 D or so.
Feed in air, at around 0.5 bars.
It´s called an air-bearing, easy to make, and *very, very* rigid, if you add suitable pressure, volume, or reduce clearance.
Someone smart might make it in 2 parts, with the reservoir inners being per-job, and the rest being at-need.
The second part has a cavity to distribute air to the ports, and a single drilled hole to an air nipple, in the center.
Bolt or glue the shoes together, yes (industrial) glue will hold fine.
Would probably take about 1 day to make a fancy-pansy great looking job with 4-6 supports, with any cnc equipment.
Or anyone with a lathe about the same 1 day to make perhaps fewer, equally-functional parts, for one who can turn some inside-radius stuff somewhat competently.
On such a big surface, the clearance can even be quite large, if you have a bigger compressor, perhaps even around maybe 0.5 - 1 mm clearance would likely work, at higher pressure/flow/noise.
(6 supports with 60 mm D surface each, makes for a lot of surface-area, and very very high support load capacity, in an air bearing).
? wont work ??
From my tlar, 6 supports of 60 mm D, at 1 bar / 0.2-0.5 mm (depending on airflow in l/min), would probably support 20-40 metric tons before contact.
Much more if the clearance is around 0.02 mm, somewhere in the ballpark near it.
(More clearance needs more airflow and a bigger compressor.)
The inner-liner holes act as dynamic reservoirs, and stiffness goes up as clearance-distance to power-of 4, while flow decreases with distance.
One could use ball-joints of any type, to mount the supports-- just use big thick ones to get the most benefit.
Ie the post below the joint must/should be thick !
If using say 60 mm D round bar to make the inner liners, I would aim for support posts of 40-80 mm D, in steel, and 40-100 mm D ball-joints.
Or just make a 2-axis clevis joint / universal joint of choice, for each support, sized for carrying 20-30 tons each.
Say 20 mm thick steel plate, 100x100 mm square platefor the joints, drill and ream holes for joint axels, TGP bar/drill rod/linear bearing rod, of about 30 mm D as the pivot pins.
Strength is not the issue, avoiding bend/jitter is the issue.
To support say 5000 kg theoretical for 50 kg practical, or 1% load, per joint.
Might cost 10-15 kg mass / joint and 40$ in materials, each, for 3-6 of, and maybe 1 day of work manually on proper equipment.
The air distributes the errors and geometrical-averages-it in a quite wonderful way.
All above is related to my spindle-building exercises, examples I have seen, and the math / examples in university industrial engineering books.
And lots of hours of learning experiences.
The advice is worth what You paid for it so far ...
and practical demos can be made available.
Multiple benefits, esp. if arrayed semi-hemispherically, for 3-4-5-6 pieces:
Self-centering,
(strongly) self-dampening,
cheap,
arbitrarily strong/stiff,
low cost to make,
low materials cost,
low skill needed to make,
relatively low work-hours needed,
arbitrarily accurate to even below sub-micron range, if needed,
( ..with some secondary reverse processing (grind the inside pads)(lap the pads).)
Ie gets stronger and better the harder You press down, a 60$ air manometer will regulate it, no electronics needed.
+Improves tir, despite spindle errors.
+Improves apparent or functional machine rigidity, perhaps 10x.
Because a lot of the stress is now spread outside the contact point, in a geometrically-increasing way, on perhaps 50x50 cm area vs a tiny contact point at the grinding wheel.
+All the bits can and will be used in the future, since You already bought a machine to do this stuff, so obviously You will encounter these issues all the time.
And the clevis/pads/joints means 95% of all work can be used with many-or-most-all new jobs.
Just make the support arms for each pad flat bars, clamped in place, from outside.
Say 40-50 mm x 80-100 mm arms, held in place at the periphery of the machine, thus nothing gets in the way.
Just my 0.03€.
If it is unclear, I can make a 3D model privately.
Take 1 day, 2 at most.