implement and benchmark C++11 std::thread vs openMP

[mikejiang]

Since openMP has some portability issue with older clang.

[mikejiang]

TBB might be the better choice, which operates with tasks, not threads. Similar to openmp, TBB's scheduler utilizes all cores by allocating a pool of threads and letting it dynamically select which tasks to run. This is the main advantage over STD:thread, with which you will need to map available work to threads manually.
It may have better cross platform support compared to openmp.
Thoughts, @evangreene?

[gfinak]

The task scheduler seems nice and tasks should be more light weight. I like the idea.

[mikejiang]

The only concern is the extra c++ library dependency (which is non header-only). We can easily bundle it,but it would be nice to able to use the existing one (RcppCore/RcppParallel#88) before we repeat the same work.