Upgrade to signal_processing 2.0 branch #77

We are currently discussing on the mailing list whether to upgrade to the 2.0 branch of the e-divisive implementation we depend on: https://github.com/mongodb/signal-processing-algorithms

I create this ticket to track the question, discussion is still ongoing. Findings so far:

• Adds two algorithms: 1) compare 2 distributions for similarity or distance, 2) outlier detection. 1) exists independently in Otava, and 2) it's unclear why or what is the use case?
• Multi-variate support in e-divisive: The algorithm supports consuming multiple metrics together, which should improve accuracy. While this would be great to have, I should not my memory of reading about this is that improvements were marginal. (But real.)
• I seems someone from MongoDB perf team might join the mailing list. It will be interesting to see what they say!