I would not say that the performance test is "recommended" at all. While it is nice to get results from the users willing to run it, I think it is too much to ask for the more common tester. The selinux testing will appropriately skip if selinux is not enforcing, and honestly the data on percentage of testers not running in enforcing mode is more interesting than the results of a test which should pass or fail depending on code, and is not hardware dependent.

Perhaps then the instructions at https://fedoraproject.org/wiki/QA:Testcase_kernel_regression should be updated to reflect your views.
The "How to Test" section is written in the imperative with no indication that the performance test is "optional" relative to default or that any variations in the SELinux state are permitted. Naively, I have always understood this to mean that there is one set of instructions from which I should not deviate.

Indeed they should. Those test cases were not written by the kernel team.

ok then - I have no idea who the responsible party is or what the more desirable set of instructions would be, so I'll be on my way...