Dave from Xylotex
I would be worried about the gantry twisting so it's not at right angles to the bed. The gantry is a long skinny triangle that spans 5 feet between the rails, yet has only a 2-foot base of the triangle that engages the main longitudinal rail. In order to deflect the gantry so it's 10 mm off square, the two sets of bearings running along the rail need to depart only 2 mm from being aligned with the rail, which could be a displacement of only 1 mm each. It's seems like that's a 10:1 mechanical disadvantage. The fact that the cross-rail of the gantry is elevated suggests that it could also be "wound up" in torsion by pressure out at the far side of the triangle, lessening stiffness further.
I haven't ever built a gantry-based device, but I've looked fairly closely at several over the years. One was a giant gantry milling machine, one a CNC wood router, and one a huge device that drew pictures by dispensing sand from a hopper. All of them had positive drive on both sides of the gantry, synchronized to keep the gantry square to the bed. With two-sided drive, any force that would tend to twist the gantry off square is immediately transferred to the racks, making it easy to keep the gantry square within the backlash of the rack drive.
With a driver like the Xylotex, it should be easy to put a stepper on each side of the gantry, engaging two racks. Use one Xylotex channel for each motor (you can't parallel the motors), but connect the STEP and DIR signals from the controller to STEP and DIR inputs for both Xylotex channels. As long as the motors don't stall, they'll stay in sync with this arrangement.