Pure TPU wheels in the rear proved too soft to do the job.  They just sucked amps & hardly went anywhere with a payload.  Making them the right hardness without an arch required a much larger hub with thinner tire.  The flat tires don't have the structural hardness of an arch, so it all comes from more material.  The new hubs would be PETG to handle the heat.  

The mane problem with a large hub is keeping the tire in place, since the tire just slides out & pushes such a large clamshell aside.  After a few failed farsteners, the best solution was to rubber cement the tire on.  Rubber cement secures it enough, the hub may end up 1 piece.

Warmer weather was causing the motors to overheat a lot more, so there was a keen interest in increasing motor torque.  Getting enough torque for such large wheels proved difficult, so there was an attempt to make the smallest possible wheels.  The smallest wheels before bottoming out were 74mm.

Helas, this proved too low for the terrain, so back to 108mm it was.  The next step was starting to passively cool the motors with heat sinks.

The only aluminum outside China was 1/32" thick.  The trick is the only way to test the heat sink is overheating the non heat sinked motor.

The next problem is the motor mounts were converted to pure PLA instead of part TPU to try to simplify fabrication.  This caused the farsteners to come loose more easily.  It seems having part TPU motor mounts either created a self locking tension on the nuts or dampened vibration.   

The way the motor mounts farsten to the chassis has always been ugly.  The farsteners are hard to reach.  They tend to loosen on their own.  It may be the farsteners just have to be bigger, but another idea is creating an all new chassis with motor mounts that attach to the angle rod.  It would just require gluing more parts & that requires a lot more infill.