optimize front() for bust of data

[azuresigma]

currently, front() always do head_.load(std::memory_order_acquire. I think it can load(std::memory_order_relaxed) first, then try to load(std::memory_order_acquire). This will help when producer has a lot of data coming in at the same time.

Facebook folly queue has a pull request for this. facebook/folly#1133. Not sure why it was not accepted into master.

[rigtorp]

Yes this is an interesting optimization.