It is good that you are open to possibilities.
Another odd thing about power that I learned was that the placement of the decoupling caps is not at all critical. This guy had designed a test board with caps at less than an inch from the test points and at about 3, 6 and 9 inches. The graphs of impedance between the planes are nearly identical regardless of which capacitor was populated. So it is clear that the caps don't need to be right on top of the power pins. But to make the smaller caps work well, you do need more of them than you need of the larger value caps. Not that I am doubting that you had enough. I am just pointing out what the data indicates is required.
I have no doubt that there are some designs that will not do well in an FPGA. These are complex chips and it has got to be hard to design them for all applications. It is also possible that there are SI issues inside the chip. As you point out, ground and power bounce are very real problems and there is nothing you can do on the outside to mitigate them.