An other thing is, that the original question still has no definite answer.
And it likely never will. It would have a definite answer on a fixed hardware platform, like the PlayStation or XBox, but not on real computers. These benchmarkexamples are not representable for practical situations? (or in all circumstances?)
They aren't really benchmarks. From wikipedia: Benchmarking is not easy and often involves several iterative rounds in order to arrive at predictable, useful conclusions. Interpretation of benchmarking data is also extraordinarily difficult. Can there be an example that is?
No. If there was, it would have been written by now. Or is it just a matter of whose opinion is most convincing?
No. It's a matter of circumstance. A greater optimization, regardless of whether one uses LUTs or FPU instructions, is optimizing your algorithm (for instance to allow for out-of-order execution, or reducing time complexity).