it should not be possible to make these for $40.

 I am considering the mill arbor (generally).

Clearly runout is bad, since runout at the arbor is going to be larger at the saw or gear cutter OD.
Well I can't measure any so I've got that going for me, which is nice.

 But-but-but, those pesky spacers!  In order to - again - not have our saw OD waving up and down, the spacer ends need to be ground flat and parallel with each other.

 On the other hand, I'm not sure concentricity to the arbor matters much at all.  These are part of a clamping mechanism.  Having a really close fit on the arbor would be nice, but the fit of the saw ID onto the arbor surely matters much more.  However, it very much looks like their OD was ground with all of them mounted on the arbor.  Not sure.

 All of them are keyed, but the last few won't pick up the key unless you put a thick one at the bottom.  The more of the stub sticking out, the harder life becomes.  Doesn't matter to me yet, as this saw is not keyed.  Set things up so that sawing tends to tighten the nut and you're golden.

 If there is a burr on the end of a spacer, it shows on the saw.  Found the ringer by checking all of them on the plate one at a time.

If there is shmutz on the end of a spacer, it shows on the saw.

 When everything has been cleaned, checked on the plate, assembled with the care of a deep space probe, and reinserted into Igor (the mill), the saw runs perfectly true.

 I have also proven I can make deep cuts with it in aluminum, despite the lack of dishing or tooth offset.  I calculate the RPM from standard SFM, then reduce it to 1/3 of that.  About 100RPM seems okay.  Patience is the order of the day.  Knock on my wooden skull.

 I really ought to fix the power table feed soon. @_@