Computation of PCA and neighbors is now part of thecs module

[MeyerBender]

Previously, in the cs module, users had to run normalization, log-transform, computation of PCA and of neighbors like this:

# normalizing and log-transforming the counts
sc.pp.normalize_total(adata)
sc.pp.log1p(adata)
# computing a PCA and neighbors
sc.pp.pca(adata)
sc.pp.neighbors(adata)

Now this is directly part of the cluster_connectedness() and silhouette_score() methods.

[MeyerBender]

Addresses point 1 of #147, still need to update docs after all of these things have been implemented.