Use multiple subscope buckets

[cdvr1993]

We have an application that has a cpu limit of 30 cores. During the start of it we see a considerable increase in latency 20ms -> 100ms. After tracing it down, we found the source of the problem in the lock contention from the subscope. We tried this change and were able to remove any latency increase.

I am also adding a parallel benchmark so that it is easier to see the lock contention.

Benchmarks Current

After the change

[CLAassistant]

All committers have signed the CLA.

[cdvr1993]

@vdarulis @mway @andrewmains12 could you take a look at this?

[cdvr1993]

@vdarulis done. The benchmark improved slightly:

Note: I changed the names of the benchmarks because they were not reflecting what was actually happening.