With this method you need to calibrate each unit, spend extra time on coding. Calibration equipment is required, ...
So then my assumption is correct, that you could use lower tolerance (read 'cheaper') components and compensate in software.
Without a micro this is not an option. I recall a blog where Dave was
ranting mentioning that he received components that were out of spec from his supplier. If the unit had a microcontroller, the errors could have been accounted for in sortware.
I agree that this calibration could be difficult. I wonder if you could build a test/calibration jig that could work with software in the PS. The PS could talk over it's serial bus telling what voltage it is sending and the test jig could tell what voltage it is getting. The PS could then log that bit of information. Do this for several voltages. Then make an other pass, this time with adjustment factors applied.
After the voltage then do current in much the same way.
I would think that this sort of calibration could be performed in about the same amount of time as a functional test would take.
I have not thought this all the way through but just off the top of my head. Therefore, there may be large problems that I have not considered.