added summary stat performance test

[ct-clmsn]

performance test added - accessing an array that stores indices, used for a 2nd matrix/array look up, could use some performance tuning in phylanx. this test provides a python/numpy version and the phylanx version.

[ct-clmsn]

this unit test represents a performance issue the white paper algorithm implementation demonstrated. it's a very simple operation but necessary for computing summary statistics.

[hkaiser]

Thanks Chris, we'll investigate.

[ct-clmsn]

@hkaiser updated with another edge case - would ya'll prefer if these tests were split out across multiple files?

[hkaiser]

@ct-clmsn I think we're still fine, thanks!

[hkaiser]

@ct-clmsn we'll have to get back to this once @rtohid and @justwagle have fixed the set primitive as currently the Phylanx frontend generates broken code for your example. I'm not saying that this is the reason for the performance you're seeing, I just would like to compare apples with apples.

[ct-clmsn]

@hkaiser no problem; thanks!

[hkaiser]

@ct-clmsn Actually, closer investigation shows, that the bad performance is caused by the (bad) PhySL code generated. Let's see how it will fare once the frontend etc. has been fixed.

[ct-clmsn]

@hkaiser apologies on my end for not taking time to vet the PhySL layer. I'll add that to my code review process.