I don't think it's actually LIMP. OP is just using that term because he doesn't know what it's doing.
The ~2 minute thing sounds like what I was dealing with in the beginning and you'd helped me with. STFT's would be fine at idle for a bit. Then they've methodically lean out. Then back to normal. Then lean out slowly. You'd said the MTF was off and it was using the failed MAF table.
I corrected the MTF and all was well.
The leaning would fit with the O2 code I'd think.
But if he'd actually pull a log and send it your way we'd know for sure. I'm just not sure why he hasn't, yet complains daily.