[FR] Aggregated statistics for all threads in a benchmark

**Is your feature request related to a problem? Please describe.**
First Issue. I would like to report the speedup (T(1) / T(n)) when using multiple threads with ThreadRange(). There exists ComputeStatistics(), but it only contains values for one specific thread. It seems there is no way to get all values for a single named benchmark. Please correct me if I´m wrong.

**Describe the solution you'd like**
Ideally, there should be a method similiar to ComputeStatistics() that aggregates all values for all threads.

**Describe alternatives you've considered**
Maybe access to the State can be provided in ComputeStatistics(). This way, at least each run can be identified by its thread_index and its name. Values can then be stored externally.