Explore usage of HdrHistogram library

[akakitani]

Explore usage of https://github.com/HdrHistogram/HdrHistogram

From the bit of reading I've done on it, that library seems to provide an accurate way to merge data from multiple histograms into a single aggregated histogram. It may be possible to integrate that with Spark 2.0's newer accumulator APIs - that would be a much nicer way to provide aggregated histograms that are more accurate and require less processing on the driver