!!! I wrote a long ramble on my thought process on a setup almost exactly like this just a few days ago... never thought to add an inductor in series, brilliant.

I can't say I understand Inductors at a theoretical level either.
My knowledge is pretty practical. For this project, I was using teensy-tiny GM-10 motors from Solarbotice. https://solarbotics.com/product/gm10/
These motors have very tiny armatures and the coils are low inductance since they don;t have very many windings.
The locked-antiphase drive mode of the H-Bridge applies an alternating positive or negative voltage to the motor armature all the time, and the frequency was low enough compared to the time constant of the armature so that the armature was able to fully respond to the voltage and draw a lot of current. This made the armatures get hot.
Adding more inductance in series with the armature made the system respond more slowly to the applied voltage, so the current built up more slowly in the armature.
The result was that as the H-Bridge switched from + to - faster than the motor could respond, the current stayed low, and the motors didn't heat up.
OpenServo also has some discussions about locked-antiphase and small motors here: http://www.openservo.com/forums/viewtopic.php?t=142

I'd love to hear more about the physics behind this. Inductors never really stuck with me from E&M.

Cool. I'd like to see more about that robot. Are you planning to post that as a project sometime?