By definition, a model is a simplified version of a system that intentionally ignores certain details and complexities to focus on key factors of interest. Not all models are created equal, but good models are useful as long as they can consistently and accurately predict results in the real world.
There is a lot of support out there for the adequacy of proper ballistics gel testing, and our resident expert says so as well.
So can someone design a product from the ground up to perform well in, or even ace a given test? I'd say yes. But if this test is a good model for bullet performance in the real world, why wouldn't you aim to ace that test?