I looked at the build thread and it looks like you've definitely covered the bases well. The primary comment would be that all of these types of motors are "intermittent" rated. If you are running them below their maximum efficiency point very much they will likely wear out fairly quickly even with good ventilation. That said, It looks like (from your other post) that these are only lasting 2-3 months? (Feb->Apr->now), and that does seem pretty short. I haven't taken mine out to inspect them since I've had them in (~4 months, 50A total limit, 22.5V max, no additional cooling), but hadn't seen anything from any other users that reported such short liftetimes. Mine get run almost exclusively on hilly pavement, and most of the time with "slicks", so overall they get pretty easy treatment. (That said, the same treatment at similar speeds torched the stock motors).
Do you have any datalogging or indication of how often it's pulling at or near the 100A limit while already moving (eg. hillclimb/etc.)? If it's often hitting that limiter then that would suggest that lower gearing may improve the situation. The thermal strain is related to the current being drawn, not the applied voltage, which varies because of back-EMF in the motor. You could reduce the gearing and raise the max voltage to make up for the speed loss. An 8514 motor should have ~140mohms of resistance, so 50A is only ~7V plus whatever back EMF the motor is already making. The motor is most efficient at 20A according the the datasheet, so even if you needed to up the max voltage you'd probably get better life overall.
Unfortunately the only stock lower gearing is an older Hurricane gearbox combined with 14T pinions (a 12% reduction). You could combine this with a correspondingly lower current limit (44A) and higher voltage (20.5V) and get the exact same max torque and max speed that you have now -- but with 20-25% lower thermal loss under high-load condition and some amount of accelerated wear due to higher RPM under lighter loads. Power loss in the windings is I^2*R, so reducing the current by 12% will reduce the thermal loss by about 25%. This would reduce brush temperatures as well as the overall current density, which I'd think this would reduce brush wear substantially if it's often running 40-50A today for any extended period of time (>5 second duration). It's just a really small motor to be sinking 300+ watts into on for any extended period of time (300W is how much it burns in the windings at 50A).
Reducing the current by itself (without changing the gearing) likely won't improve the situation as it will just sit on the limiter that much more often (running slower, for longer duration).
I'm not sure whether that would end up being a good tradeoff or not without having some sort of datalogging of the current being used during regular driving now.
You probably won't have room to do what i did on my son's F150 truck -- which is to install two 775-series motors in each gearbox (posted under "quad damage") to help spread the thermal load.
I haven't see anyone install a CIM motor in the stock gearbox -- that would be an interesting build. That's an example of a motor with substantially more heatsinking (in the form of mass, mostly) -- shows up very clearly in the extended power tests. I'll post it here since they also test the Banebots 18V / 8514 motor. Vex motor comparison
F-150 - 24-36V with homebrew 100A variable-speed motor controller, 4x 775 motors (NOT all-wheel-drive), and fading color
Hurricane - 24V w/50A step-down controller, Mabuchi 8514 motors
Princess Mustang - 12V w/23000 RPM 550 motor and 23T gears