Spent some time with this and a dial indicator to try and figure out what in the world is going on.
Come to find out it was mechanical, not electrical. The Y-Axis (bed) rails were badly worn (7.92mm, way out of spec..) and the original pillow blocks & dual LM8UU bearing config wasn't doing a good job of keeping things from wiggling around.
I ended up replacing the rods with proper hardened and ground linear rods, as well as redesigning the pillow blocks to use an LM8LUU paired with an LM8UU for better support. This fixed the mechanical wiggle, but the weight of the bed applying cantilever force to the pillow blocks was still moving things slightly, and the Y-Axis difference between the probe and nozzle were making this impossible to account for.
So, I redesigned my carriage assembly to put the BLTouch directly inline with the nozzle, no more Y-Axis offset.
fbc1d4da-8a49-4bd3-889a-f6f11c20b153-image.png
And here's the probemap.
b213c080-7b64-4141-b8a3-89f2f3160ce0-image.png
I've now successfully printed several test objects around the max size of the bed without issue.
Big thanks to @dc42 his help, and for getting me thinking if those trigger heights were real or not. That got me down the path of challenging the mechanical on this.