Did the best I could to characterize the spindle bore, and was pleasantly surprised. Really need a profilometer to accurately assess the ridges, but using a Tesa 0.0001" indicator with my compound set at 1.5°, I scanned the MT5 bore as far as I could with the compound (~1"). I was getting fluctuations of <+/- 0.0001" as I scanned from out to in the flat portion of the taper at 4 different radial positions about 90° from each other. It looks rougher than that, and if I had a stylus instead of a ball I would read greater fluctuations, but the ball should be close to what the 5C adapter "sees". I checked run out in a few different distances from the nose and the indicator barely moved as the spindle turned--maybe 0.00001"?
I cleaned the spindle bore and collet adapter a few times with WD40 until the paper towel didn't remove any black stuff. I slid the collet adapter into the spindle and gave it a thump with the heel of my hand. I was pretty happy with how firmly it "stuck" which I take as another positive indicator.
I indicated a few different distances from the nose of the adapter and again got maybe 0.0001" to 0.0002" total movement of the indicator.
I used the collet closer tube that came with the lathe to try to manually close a 1/2" collet on a 1/2" drill. It spun the collet adapter in the spindle nose. It's pretty clear now how the spindle got scored.
I decided to try out the pneumatic collet closer that came with the lathe. It's a bit leaky, but seems to basically work and clamps the collet, firmly holding the 1/2" drill. Rotating the spindle with the indicator on the drill shaft I'm seeing about 0.0005" - 0.0007" fluctuation. This seems decent, although it's only about 1/4" from the collet nose, so runout would obviously be a lot worse further from the nose.
Gonna need to make a few parts for the collet closer. This lock ring that secures the bearing for the collet closer is pretty messed up, both the cavity for the spanner and the threads.