Thread pools were devised to eliminate/mitigate the overhead involved in thread creation. But I'm having trouble proving to myself that (given some minimally nontrivial work) a thread's creation time ...