I like that all important points regarding using differentials had been covered already:
- every additional axle introduces extra friction losses - no way around it
- it is possible, but hard, to maintain high build quality to manage friction losses
- since systems are linked, it requires constant preventive maintenance to be reliable
Four bar could be made to work, but you pay the price with additional friction losses. My team had 4-bar differential in ITZ with 4 motors on 12T-84T-84T-12T geartrain connected to other 4 motors direct driving the wheels. It was ok, but it was, probably, losing at least one 393 motor worth of power to extra friction. With lower V5 motor limit that would be too expensive, so it made sense to switch from 4-bar to coaxial type differential to reduce friction. Similarly, this year coaxial 4 motor drive to tilter differential had been performing very well. It is the rest of the robot that needs more work and attention, because getting differential to work well and programmed correctly takes time.
There are only two scenarios where it makes sense to use differential. First one is where you have no other way to power all functions on the robot without reducing drivetrain power. However, you have to make very strong case why that design could be higher performing robot than well built and tuned Goofy with driving team having few months of practice.
And the second scenario is when you want to take the harder route to learn about system reliability, advanced mechanics and programming PID for complex systems with multiple degrees of freedom.
You may find yourself spending most of the time improving your build methods and doing in-depth research on how PID works at the expense of more driving practice. It all depends on the team which of those could have greater educational impact.