stoichiometric is 14.7
Everyting below that is theoretically rich. Until 2600 its about 14, so what is 'lean'
But it almost can't be a dyno problem, because otherwhise the torque and power curve would gradually fall off, and instead it rises faster than with the rich mixture.
If it would be a dyno problem both the afr sensor/calculation (I don't know; do they pick that up directly from the first lambda probe?) and the power sensor/calculation would be at fault. That would be a coincidence...