Actually your open-loop MAF tune was not that far off. You must have got it close when you did it via the NB's.
I usually just use the 'canned' BEN Factor 1, Serial Wideband LS1..See Attachment.You have to have each of those PIDS in the equation selected to have a CALC.Pid work.
So your Tune experiment was actually pretty close. I would still try to do it using both NB's and WO2. 12.8 AFR vs 13.0 is within 1-2%, and even your closed-loop AFR's were pretty decent..