Same thoughts as Bolt did occur to me but I forgot to mention the potential for a gearbox problem being the underlying problem.
I do think though, like Gilbert, that the problem is probably due to duff lambda sensors. I have seen a lot of P38s with duff lambda sensors usually with the owner and general garages the owner used the services of unaware.
Normally if a 0>1v lambda sensor fails it will output close to 0v, which for a 0>1v sensor implies lean mixture, but the error codes point to rich mixture while the symptoms point to lean mixture. If lambda sensors incorrectly read rich mixture it may have the effect of leaning the mixture off. Given the aforementioned (question to everyone) - What are the chances that someone spliced in 5>0v sensors instead of the 0>1v sensors? I know that in this company I have no need to mention that with the 5>0v sensors a 5v signal; points to lean and a 0v signal points to rich...So if a fitter had mixed up their P38 models and spliced in 5>0v sensors on a model that should use 0>1v sensors there would be constant rich running error codes and the fuel trims might be leaned off leading to reduced engine torque, the engine would still rev but the car would lack power and go up through the gears slowly.
I currently have @Jacckk 's P38 in the yard, the last time I saw this P38 was 10 years ago but he fitted a 4.6 to replace his 4.0 and the LPG fuelling was off after the engine change. Today I got it running much nicer on LPG than on petrol, probably due to some petrol injector issues (only petrol injectiors on 7 and 8 seem to be working great, others probably not working very well).