If that were the case, that would explain a lot. I noticed that the Z-Axis wasn’t moving as much as it should have during the 27 square test. When retracting, it would slowly turn (like, very much slower than usual), or not at all. I had thought that maybe a wire had gotten disconnected, but it also would be entirely possible that the router was getting hung up on the Z-Axis motor. I looked through the video I took yesterday but the motor and leadscrew weren’t lit well enough for me to really see what was happening.
I saw somewhere around here a mod where someone connected a piece of plywood to one of the base’s guide holes and used that to mount the Z-Axis bracket. If I do settle on the 45 degree system, I’ll have to do something like that to align the motor better.
I know, I wish I had enough time to do all of them back-to-back. I might be able to sneak in some time during the week, but I’d probably be able to do only 1 step of the test than all 3 at once.
I have the first version of the top mount:
I have noticed a fair amount of slop in the holes. I’ve been trying a couple different hardware options to see if I could eliminate that, but it’s good to know that the holes are tighter on the current version. I’ll gladly test anything you send me 
Thanks! I don’t have a scale big and precise enough to weigh the entire sled at once, but I have weighed out all of the components:
I put the total in the main bed accuracy test spreadsheet, but I’ve only been posting screencaps of the relevant data. Here’s the spreadsheet I’ve been working on, in case you’re curious. The header has all the basic calibration values in it.
I might have messed up the chain sag calculations, though, now that I think about it. I only did what was in the auto calibration dialog, and don’t remember entering sled weight or bed angle into Ground Control. That could certainly explain why things were off in the center.
As far as the motors flexing, I haven’t tested or observed any flex in their brackets, so I’m not sure. I’ve never seen any noticeable deflection, but even a slight amount could have an effect on the tests. I had to make an adapter bracket from 1 1/2" x 1 1/2" x 1/8" steel angle to get some height out of the motors. Admittedly, they don’t even give me enough height. I plan on replacing them with a Unistrut top beam, but if the little guys are flexing then they might be forcing my hand on this.
Pictures of the assembly, both with and without the chain guide. The chain guide mounts to a second piece of steel angle.
Next time I’m in the shop, I’ll have to aim my camera at the motors, then raise and lower the sled the entire height of the bed and see if we can notice any deflection 
If I’m geting flex at that point, does that invalidate a lot of the data I’m collecting? Really hoping I just messed up the math 