They had to swap out the auto/manual MDL after OFT-2 for CFT-1, that's apparently the designed mechanism for making the mode change.
There's a lot of configuration management activity behind doing this they haven't described, but having worked with 3 other aerospace contractors beside Boeing on their flight software management, none of them would exercise any less diligence in configuration-managing such a change to insure it was made reliably. Note, NASA wouldn't accept any less, either.
What was really the issue is that they chose not to run the software qualification campaign for the CFT-1 software version with both auto and manual MDLs, the reason for the 4-week estimate to revert to auto mode in the in-flight vehicle. However, if they're planning to return the vehicle next week, they were apparently able to do the re-qual testing in less time, yay.
