add action supporting generation of randomized metadata groups

[gregcaporaso]

Resolves #47

This allows users to create one or more ~randomized~ shuffled categorical sample metadata columns, with the number of groups and the count of samples assigned to each group matching that of the input metadata column.

An example of how this can be used with the Moving Pictures data is as follows.

qiime metadata shuffle-groups --m-metadata-file sample-metadata.tsv --m-metadata-column body-site --o-shuffled-groups rand-md.qza

qiime diversity alpha-group-significance --i-alpha-diversity core-metrics-results/faith_pd_vector.qza --m-metadata-file rand-md.qza sample-metadata.tsv --o-visualization core-metrics-results/ags.qzv

---

First, alpha-group-significance on the body-site column:

---

Next, on a randomized metadata column modeled on the body-site column:

[gregcaporaso]

On it.