Since this is a design goal of GaudiHive/AthenaMT, adding a test to ensure that algorithms can actually use internal multithreading via TBB.
At the moment just uses TBB parallel_for. Could add a more comprehensive test with task_group if that would be valuable?