I just wanted to update the thread. I checked run-out on some other machines and decided I had room for improvement. My spindle has 2 opposing tapered rollers spaced about 4" apart on the nose end, and a third support bearing on the 12" long spindle extension exiting the back of the headstock. For the sake of simplicity, all measurements were made with the support bearing REMOVED, so it didn't influence or mask the run-out in the main bearing set. Before the rebuild, I had .0004 Tir on the nose of the spindle taper, that included ~.00015 of run-out that only repeated every second revolution. The end of the spindle extension had .0015 TIR, with .0005 non-repeating run-out.
The old bearings I removed had a lot of radial scoring, but I didn't see any indentations from debris. They were standard bearings, not precision class. I replaced them with identical off the shelf SKF bearings keeping everything as clean as I could, and set the preload to 15"lbs of rotational drag.
After the replacement, I had .0007 TIR on the spindle nose, but I couldn't detect any NON-repeating run-out with a tenths indicator. On the extension, I had .002 TIR, again with .0000 NON-repeating run-out. In other words, the indicator comes back to the same spot every rotation. So overall run-out on the spindle went up, but the NON-repeating run-out that was causing me problems dialing parts in is essentially gone, or at least undetectable with my measuring equipment. Thanks for all the help and insight guys, I wish my TIR would have been less after the rebuild, but since I never use anything mounted directly in the spindles taper, I'm satisfied with the results.