* add content-base features to user features (#1)

* add user tag feature

* add precision at k metric

* some tuning

* readability fix

* small updates (#2)

* comment update

* update sql docstring

* change precision to recall

* fix bug when num_users or num_items is too small (#3)

* update typical users

* small fixes

* fix bug if dataset is smaller than expected

* use correct item and user sample sizes

* have recall at k handle out of range ks

* split user tag steps and add unit tests

* update readme

* missed docstring