You might find the afr variation is down to changes in IAT. Hence why using that mod table against injector flow might help you reduce that error. It used to on the old LS1 setups with the COS. Log BEN's vs IAT and see if you see pattern, if so try that.
Personally once you are close Id just to a long drive like Swing says and sort out the VVE, then consider it done and look elsewhere to clean up the variation. Otherwise you can end up chasing your tail all the time. I doubt you will ever get it to track 100% all the time.