Revert Qthreads patch from #26328

[bradcray]

This task is a placeholder to mark our intention to revert the Qthreads patch added in #26328.

Though that PR seems to have helped with the performance of the test it was designed for ("Array init/deinit performance, 40000000 element array"), it also hurt the performance of several other, arguably more fundamental, tests (link to graphs).

If there was a strong sense that the patch was the most correct and appropriate thing to do, we might keep it and work through the backslides, but talking to @insertinterestingnamehere , it sounds to me more like it was an approach that seemed to be working for the test we were focused on, but not a change with a strong rationale outside of that.

As a result, my current expectation is that we'll revert it in the new year (assuming nobody gets to it today), and I'm filing this issue to remember that.