Balancing sounds like a good idea and is best done at some known reference level, either at the top or the bottom end of the cell voltage range.
The question here is what has caused the imbalance situation with the two cells--Were they drained from providing power to the BMS board, or do they have an internal defect? Are the two cells neighbors or is there good cell(s) between them?
The balancing in the miev is only a bleeder function at a very low current, so it would take a very long time to correct even a small imbalance voltage.
It somewhat depends upon how much work you want to do, e.g. pull and open the pack several times vs one time, open the pack after fully charging or fully discharging, etc.
Some scenarios,
Open pack and use lab power supply to charge just the 2 low cells up to match the rest, then assemble and test.
Open pack and replace the 2 cells after bench charging the new ones up to match the rest, assemble and test.
Open pack after being discharged, use lab supply to charge 2 low cells up to the rest, then discharge entire pack toward the bottom while making adjustments to the 2 bad cells to prevent going too low, etc. with goal to bottom balance the entire pack (with the laggard cells). Once balanced, assemble and test, this might work if the cells were just drained from the BMS, but not if they were defective.
Same as above but done at top voltage, top balancing.
Open pack, remove the 4-cell module and swap the CMU board between the old and new module, charge the new module using bench supply to match the rest of the pack cell voltage, assemble and test.
Same as above but fiddle around trying to remove and swap eeprom chips, or reprogram eeprom chip.
etc...i'm sure there are other options too.
