Draft: Cohorts model

Since theres no easy way to calculate cohort data on the fly using Superset and our current cohort models are specific to a single set of use cases, I created a new cohorts model that is generic and can be filtered to see pre-aggregated cohorted data. We'll still need to add them manually to this model but we wont need to import new columns or customize visualizations (other than switching some filters).

• Add a generic cohort table
• Calculates activation date for projects, this allows for more accurate understanding of when events start contributing to the metrics that care about active vs inactive
• The current set of cohorts in this model are: event_volume (month), active_projects (month), APP_active (week/month), APP_active_prod (week), APP_active_dev (week)