Mate, I think it's time to go down this road.
That printer is so wonky, I think you've got to get into its firmware. If you can get Marlin onto it (as per that guide), and it's not a genuine hardware fault, you *will* be able to get it working. Set aside a full afternoon to bootstrap it, though.
Right now, you're Father Ted trying to knock the dent out his borrowed motor!
After that ask it to print thin, flat rectangle 3/4 the size of the plate and get a-measurin'. A long baseline needed to work out steps per mm.
As a ferinstance and not directly related to the problem you're seeing, my (shitty cheap) i3 clone has a problem where the left and right y-axis bars aren't quite straight to each other. As the print head moves in +x, the whole mechanism tilts forwards. Because the z-probe is mounted at the front of the carriage, as it droops forwards, it thinks the bed is getting much closer than it is to the nozzle (the bed is getting closer, but it's farther up the lever arm -- ie. it overcompensates).
This would be a nightmare to fix physically. It's probably just a slight warp in a couple of parts of the frame, but which ones and how to remedy it?
But with access to the firmware I can just add a linear z-probe offset fudge equation, dependent on the value of x. Job's a good one.
Use the (open) source, Luke!
BW