RJ,
As I don't do this for a living, I thought an error of about .001% over a span of 21 inches would be considered excellent. That being said, after reading your post, I called one of my friends who runs a small shop that specialized in stainless steel and other more exotic metal repair and parts for the local drug companies. He said the calibration company, that comes twice a year to his shop, runs a somewhat similar sounding tests, 5 times on each machine and get an average travel error. He said the 70s era Wells-Index (completely rebuilt in 2012) is off about .045" over 36" or about .001% and the 3 year old Bridgeport, has an error of about .031" over 36", or about .0009%.
As I didn't witness the OPs test or witness the calibration procedure, that the friend said they seemed very similar, I can't say it is, but according to him, he didn't think it was excessive error. I'm not trying to be argumentative, I'm looking for information. Obviously I need to do some more reading on the subject.