Make treespace and associated plots N-dimensional

[danlwarren]

Right now it's hard-coded at 2, but it's easy to make this stuff N-dimensional. May have downstream impacts on other plots, though, which we'll want to take a hard look at.

[danlwarren]

Another thing I think would be really, really neat would be to have something where we could correlate clade presence/absence with each of those dimensions and then plot an N-dimensional treespace above the diagonal and small phylogenies ON the diagonal with the clades highlighted where presence/absence was most correlated with that NMDS axis.

[danlwarren]

Following on that last bit: having small trees on the diagonal is probably going to be completely unreadable for any large topology, so (as the product of a conversation with Rob) maybe the thing to do is make the pairs plot for topology and then find a way to print the axis loadings on an MCC tree, which we would just return as a separate plot.