Tool Holding

[Carm]

You asked what is typical for tool runout. Hard to answer with a number, but in a typical jobber or manufacturing plant there's very little that's perfect. A machinist (as opposed to a line worker) probably knows what it is and also has strategies to work around it. You have to produce parts in tolerance, do or leave. Management usually won't spend money on repairs until there's a big wrench in the works.
I'm not familiar with Erickson QC, but after pondering your conditions, think that inking (bluing) some collets while holding appropriate shanks might reveal the problem. I suspect one of the angles isn't seating fully, the DA collets take a good bit of force. If you can find one collet that suits, compare the others in all dimensions and reason carefully.
I wouldn't touch a tool or grinder to anything 'til sure of WHY and WHERE.
The more items (shanks, sleeves, toolholders etc.) added to a spindle the more expected runout but for a low RPM mill or drill .003" TIR with the tooling would be my limit. No angular.

[ctwo]

Mr. Dremel was going to be my strategy to address this. I have not measured every collet in the line. They are new and should be high precision. The nut is now new as well, yet gives the same result as the old one. The new nut feels like it locks harder, less turn after it seats...

I have blued (sharpie) the three angles and all blue is gone after seating, even just bare hand tight.

I have measured the taper at the inside end of the holder with a DTI and it does run out 4 tenths. I have yet to reach the inside taper at the end of the hole. The tapers, which are very small anyway, look like they are well used is all. I have not measured the inside taper on the nut.

Maybe I will put a piece of tin foil on the high side of the collet and see what happens. I also thought of painting half diameter of the mills and set them in a similar way to the holder.