The short answer is that you don't want to mess with the relative masses of the props because they spin opposite directions to counteract each other's rotational inertia; it's the reason these quads don't have tail rotors. Furthermore they actually use that inertia for yaw control by varying the relative RPM of CW and CCW sets.
Think about that for a second... there is no air being pushed to spin your craft around the yaw axis (i.e. panning); it's all done with inertial offset. Now watch how authoritatively your quad pirouettes with left stick input and you'll get an idea how big a deal this is on a multirotor.
Upsetting that rotational weight distribution is a great way to inherently unbalance the system, hamstring the flight controller, and generally invite all kinds of serious trouble.
Now thrust, strictly speaking, should be identical given the matching shape (length, profile, pitch) of the prop...acceleration differences due to mass/distribution are probably there but I'd be surprised if you could measure it even on a test jig....that'd be the sort of thing most observable through battery life, as it aggregates minute differences in work the motors are doing over the course of an entire flight. I honestly run out of aeronautical knowledge before commenting on flexibility...maybe someone else can.