I really like this drivetrain, like a lot. Great job! It's been a lot of fun watching it get better and better. That being said, I think there's a few things left that could be optimized.
The first thing is space. I think electronic placement space is a very important aspect of a drivetrain. Without enough space you have to spend time coming up with ways to get everything to fit, and some of the solutions can make maintenance difficult. YMMV and I know a lot of teams are willing to make space sacrifices that make electronics placement difficult, and sometimes that trade off pays off. There's nothing wrong with finding sneaky ways to fit all your electronics on the bot, but I consider an ideal drivetrain one that keeps electronics placement simple. For that reason I suggest space as something to try to optimize.
The reason I think this drivetrains electronics space could be optimized is because of the large voids in between the modules and the voids created by the hexagonal frame perimeter. I think hex and octo frames are great, but I also think there's trade offs that should be considered. The first is space, is having such a wide hexagonal robot worth it if it limits space? And how will the frame shape effect superstructure and manipulation design? You may have accepted these trade offs, but what is brought into question is how hexagonal should you make your robot given those trade offs? Looking at your frame I notice that your sides are steeper than most octo and hex framed bots I've seen. Given that there's trade offs to having a non-rectangular frame, the key in designing this type of drivetrain is to balance those trade offs with the benefits of an octo or hex to get a shape that is effective in terms of space and interactions with other robots. Hex and octo bots are pretty new in frc, so not much is know as to how design one with the right shape. To determine how steep to make your corners I suggest building bumpers of different angles and testing it's effects on robot interaction to determine the best shape. Once you've done that you'll know how important frame shape is and then you can determine if having such a wide hexagon is worth the sacrifices when compared to a slightly smaller hexagon. There's teams that have done testing on this very subject, i'm hoping someone chimes in.
The first thing I thought of when I saw this design was that you could save quite a bit of space by putting the motors in the void between the modules. I read the previous thread and I noticed you had the same idea but didn't pursue it because it would be too complex. I don't know where you or your team draws the line for complexity, but I encourage you to not give up on that idea just yet. I think there's ways of getting a gearbox and motors in that space that are a little more complex than your current design but could be a great improvement and really take it to the next level. The most obvious reason to move the motors is so they don't take up valuable electronic space, but I think by moving the motors you can actually reduce the width of each side of your drivetrain pretty significantly. When I look at your power train I see two things that make each side wider than it needs to be: the gears for driving the wheels and the pulleys that connect the omni wheels together. My suggestion for narrowing your sides is to take the whole power train and put it in the gap between the outside of the hex frame and the inner yellow frame rail. In order to pull this off you would need to come up with a clever way to either replace each motor or remove the entire gearbox, which I think can be done. The advantage of this is that by putting a gearbox in between each module you eliminate the need for a belt to connect them, making the modules narrower. The main idea behind this is to take everything that's making the sides thick and put it in the void where there is space. There's a lot of ways you could do this without making it too complex, I've got some ideas but I'd like to see what you come up with
There's some other things, I ran out of time to post everything, I might post more later.
Oh, and just fyi: as for using loctite for holding cim screws in using the strongest loctite you have is not the way to go. Look up what strength thread locker you need for the fastener you're using and use that. If you put the wrong strength loctite on you might end up with an irremovable screw.
If you're seeing unusual stress risers in your results the first thing you should do is check your mesh. (assuming that's applicable to whatever program you're using). Also, for accurate results make sure you have at least three layers of mesh in the cross section.
-Adrian