Added seriesMedian() and seriesPercentile() statistics to the series obj...

[broxtronix]

This is a pretty straight-forward addition of seriesMedian() and seriesPercentile() to the list of statistics that can be computed for a series object. These are useful for all sorts of things, but presently I am hoping to use them to stretch the contrast of the data before colormapping it, retaining only the data that falls between the 1% and 99% percentile, and clipping off the few outliers outside of that!

The median (but not the percentile) is also computed when you call the seriesStats() function, since computing the median is not a terribly expensive operation compared the communication overhead involved in getting the results back with Spark.

The percentile is only available as a seriesPercentile() method, and not as part of the seriesStats(), since it requires an argument (the percentile) to work. If desired, we could add some pre-computed percentiles (1%, 99%) to the seriesStats() method, but I think its fine to let the user call seriesPercentile() for now.

[industrial-sloth]

This looks pretty solid to me. I think I agree about leaving seriesPercentile() out of seriesStats() - it's not obvious what the default arguments there should be (though 1% and 99% does seem as good a choice as anything).

@broxtronix, any chance you could add calls to seriesMedian() and seriesPercentile() alongside the existing assertions in test_series.py TestSeriesMethods.test_series_stats()? And for seriesPercentile, could you also add a call that runs the multiple-argument version (which I think would be something like data.seriesPercentile((20,80)), if I'm interpreting the docs correctly). Thanks!

[industrial-sloth]

Never mind about the unit tests @broxtronix - I'm in the process of adding / updating a bunch of unit tests anyway, so I'll get that put in later on today. cheers...

[broxtronix]

Hey @industrial-sloth! Very sorry for the delay on checking back in on this pull request. During the past couple of days I was pulled away from coding to prepares some other work. Huge thanks for adding these unit tests, and for pulling in this pull request. Much appreciated!!

[industrial-sloth]

Not a problem at all - am going through unit tests this afternoon anyway, so it's very easy for me to get this knocked out as well. Thanks for the PR!