[FEATURE] Investigate Using a Lock Free Queue

[ptsouchlos]

I'm curious to see if a lock-free queue significantly improves performance for the thread pool. We should do some tests with industry standard lock free queues in combination with std::move_only_function to see the difference in performance.

Queues/deques to test:

• https://github.com/ConorWilliams/ConcurrentDeque
• https://github.com/cameron314/concurrentqueue

[ptsouchlos]

https://github.com/erez-strauss/lockfree_mpmc_queue

[ptsouchlos]

Got an implementation in #54 but having trouble addressing some pesky bugs as I can't pin down the reason for them. Could use some assistance.