Fuel trims work like this... An ECU has a base map/table (picture a spreadsheet) that might have rpm on one axis and airflow or manifold pressure or throttle position on another axis, numbers in table cells tell the ECU how long to pulse injectors. That's the basis of an open loop system (one that doesn't have lambda sensors). A closed loop system (that has lambda sensors) has the same but also has a table for long term fuel trims and a momentary stored value (not a table) for short term fuel trims. The injectors are pulsed for whatever length the base table says + long term trim% + short term trim%. If the lambda sensors say the engine is running lean the value of STFT increases, if the lambda sensors say the engine is running rich the STFT decreases. If the momentary STFT remains positive for long enough it starts to increase the stored LTFT, if the momentary STFT remains negative for long enough it starts to decrease the stored LTFT. Whenever you change throttle position or rpms change the new length of time to pulse injectors for is read from the base table, the LTFT is also looked up from its table and added to the base table value but the STFT is reset to zero but the STFT will never stay at the same value for long.
You can imagine that when the vehicle is first built it has it's base map but the stored LTFT's are all zero. At this point at (say) 2000 rpm with (say) 10grams per second airflow the mixture won't be exactly correct... let's say it's 8% lean. So the STFT's very soon increase to +8% to compensate and bring the mixture correct. Now since the STFT's are hanging at +8% positive they start to steer the stored LTFT's, first maybe by just +1%, so now we have the base map figure +1% LTFT and if we stayed with +8% STFT the mixture would be 1% too rich so the STFT decreases to +7%. Then it would follow LTFT became +2% and STFT +6%, until eventually we end up with LTFT at +8% and STFT at 0%. Then as we drive around we change throttle position and rpm, but when we next use 2000rpm with 10grams per second airflow the stored LTFT for those conditions will already be at +8% so now the mixture is correct without the STFT needing to adjust mixture by very much at all. Eventually when all LTFT's have been learned in this way we can vary throttle position and rpm and mixture will be very close to correct all the time with very little need for adjustments from STFT's and this helps to improve driveability and mpg while decreasing emissions.
The Ford/Jag derived engines are a bit more complicated because they also adjust injector pulse length for fuel pressure which they read from the electronic fuel pressure sensor, and they can and do change fuel for different conditions.
They are fitted with wide band lambda sensors. It is difficult to interpret a lean or rich reading from a wide band lambda sensor using just read voltage or amp readings, it is better and more accurate to read 'lambda equivalent value' (or whatever terminology the scan tool or OBS system uses) using an OBD scan tool. A lambda equivalent value of 1 is correct (stochiometric) mixture, below 1 is rich (e.g 0.9 would be around 10% rich), above 1 is lean. Or some ECU's / scan tools will show mixture in terms of mass ratio in which case 14.7:1 would be correct, below 14.7:1 rich and above 14.7:1 lean. But you don't really need to see / read lamba values for purpose of diagnosing the problems you're having because it seems the lambda sensors are working and because you can get the info you need about mixture from the fuel trims.
If LTFT's are as high as 19% with STFT's +-7% it's not too surprising error codes pointing to 'system too lean' are occurring. The +19% will be close to the maximum LTFT's are allowed to be steered to, so if STFT is positive when LTFT is +19% we might expect a lean error code.
If the trims are highly positive around idle but decrease with increasing engine load it could point to a vacuum leak, because vacuum leaks have most effect when there is most difference between atmospheric and manifold pressure. If the trims are consistently highly positive it points away from a vacuum leak for the same reason, but might point more toward a MAF generally under-reading airflow problem. But like I said fuel pressure can also be a problem on these engines.
Incidentally what fuel are you using in it? Ethanol has a different stochiometric ratio to petrol, a higher percentage of ethanol in petroll naturally causes higher fuel trims than neat or low enthanol petrol.
Old school narrow band lambda sensors like 0>1V sensors are quite generic, in most cases you can use any sensor of the same type in any car if you cut the connector off and splice the wires in. But wide band sensors are different, you need the exact spec sensors to suit the car.
I believe there have been a lot of knock-off dodgy or slightly incompatible MAF sensors for these cars.
At this point I would recommend following my original advice to clear fuel trims, let the engine cool fully down, start it and immediately set off driving without letting it idle before setting off for the drive. If your scan tool doesn't have a function to clear fuel trims you can do it anyway by causing a 'very hard' error such as causing a misfire with a disconnected fuel injector, i.e.unplug a fuel injector and drive it (it is difficult to access injector plugs on a 4.2SC, but you could snip a wire at the injector multiplug at the rear end of a fuel rail), then reconnect the injector and when you clear the error codes it will also reset the fuel trims. During this driving don't stick to a certain throttle position and rpm for more than (say) 20 seconds, keep slightly changing throttle position or rpms. This is because during this time since LTFT's start at zero STFT's might need to be highly positive or negative, the STFT's are also limited in extent of range they can be steered to and if they reach close to the maximum or minimum instead of the LTFT's being steered the ECU sees it as an error condition and won't steer the LTFT's so you get a mixture related error code anyway. But by continually adjusting throttle / rpm the STFT is reset to zero and the 'clock' for timeout for rich/lean mixture is reset, while before STFT reaches maximum/minimum it will steer LTFT's. I also believe the Ford/Jag engine related ECU's have an underlying 'master' fuel trim... If the old skool model for a closed loop system was base table + LTFT + STFT the Ford/Jag wideband systems would be base table + master + fuel pressure + LTFT +STFT with 'master' kind of being learned one time during initial running when the vehicle was new or after resetting fuel trims (or after clearing a hard error which invokes clearing fuel trims), if the master is wrong fuel trims might never work within range and without having mixture related error codes occur.
