Sort anomalies for each nightly

[shssoichiro]

Within a nightly, the anomalies currently don't appear to be sorted in any particular order. I think it would make sense to sort by number of standard deviations (most significant regressions at the top) to make it easier to see likely regressions in the list.

[anp]

I believe they're sorted by the KDE score which represents a combination of factors, but I'll have to double check later. I didn't expose those anomaly indices in the UI because I wasn't sure how to explain them, but that could be something worth revisiting.