Add Watterson's theta and Tajima's D into pixy

[npb596]

Hi Kieran!

Figured it's worth making this an official pull request by this point. Feel free to test or critique as needed.

Cheers,
Nick

[taprs]

Cool to see it implemented Nick! Afaics these D values will be shifting depending on the amount of missing data unlike pi and Dxy, right? Should there maybe be a warning about that, e.g. to discourage comparisons of window D values in presence of missing data?

[npb596]

Thanks Nikita! Yes to the shifting, and I've seen in my simulations my corrections for Watterson's theta and D are both still biased somewhat by missing data. I figured it's worth pushing this finally since it's still an improvement over several other tools. We're working on a manuscript that goes over caveats but I agree once this is implemented in the main pixy distro the documentation and code help messages should explicitly mention these issues too.