I could be too stuck on the original feeding out chains the same distance to attach the sled. This calibration approach might be able to handle that you put the chains on the sprocket eyeballing, attaching the sled and the cutting of the squares a few times will take it from there.
My square is perfect. Its just that the square is rotated a bit anticlockwise
I’ve digged in my archives, and found 2 more approaches to get the motors and sheet level.
One 2M level in the middle, 2 shorter ones at the side.

A leveller with laser, need a level on top to get the laser level.
My last approach was the clear tube with water mentioned above.
Great, thanks for this. This will then be the start point for today
In the first picture you can see the ‘aquarium pump tube’ already hanging there to be filled with water, because the outcome of 3 levels was not conclusive. Due to the small diameter of the tube, I had to beat every air-bubble out. With a bigger diameter clear tube the air will go by itself.
Edit: if it is not the level, take a look at this: https://github.com/ldocull/MaslowDue#machine-home--home-h
Edit2: so if this works, centre your sheet in the middle of the motors, mark the middle of the sheet, move your sled there and set home. The setting the sprockets on 12 o’clock and feeding out the same length of chain seems to be obsolete with this approach.

