It's pretty hard for me to imagine multiple cap failures of that nature. I might buy one on a board...
This is the study in the video: MLCC-FailureModeStudy-032012.pdf
I'd eliminate the first two categories right up front, obviously the design isn't that bad as thousands of these boards have been shipped and are fully functional.
Before I'd suspect the board, I'd want to see how this board reacts in a different environment. Obviously, I'd put it on the test stand and see how it functioned, but swapping it to another engine would be a useful exercise since I doubt you have a PS/3 board tester. I'd also be looking very closely at the rest of the engine, and perhaps doing things like checking the motor current for each of the motors using a DC bench supply.
