Putting aside the fact that I don't think you're actually disagreeing with the parent, I don't think it's really accurate to say that all programming level tech

Putting aside the fact that I don't think you're actually disagreeing with the parent, I don't think it's really accurate to say that all programming level techniques of isolation are defeatable. The paper enumerates a specific set of vulnerable language features (e.g. "Indexed data structures with dynamic bounds checks"). The paper doesn't say much about languages lacking these features. Also, I believe retpoline is a (nearly?) perfect software fix for Intel's microarchitecture.