Create popularity_refresh DAG factory

[stacimc]

Description

Create a popularity_refresh_dag_factory similar to the DAG factories for provider and data refresh DAGs. For each media_type, it should generate a <media_type>_popularity_refresh DAG which does the following:

1. Update the <media>_popularity_metrics table to include any newly added
   metrics.
2. Refresh the <media>_popularity_constants view to recalculate the popularity
   constants.
   1. This is done CONCURRENTLY so that provider DAGs can continue reading
      from the view while it updates.
3. For each unique provider in the <media>_popularity_constants view,
   generate a refresh_<provider>_scores task. The task will run an UPDATE of
   the standardized_popularity on all records matching that provider which
   were last updated before the task began.
   1. We may consider running the refresh_<provider>_scores tasks in parallel
      to speed up the update.
   2. Optionally, we can hard code a SKIPLIST of providers that are present in
      the <media>_popularity_constants view, but for which we do not want to
      create a refresh task. We currently have some providers (Nappy, Rawpixel,
      Stocksnap) that support popularity data but are not dated, meaning scores
      for all of their records will be updated the next time the DAG runs.
      Note that some of these DAGs are on a @monthly schedule however, which
      means skipping them in this DAG could result in delayed recalculation
      time.
4. Report to Slack when the scores have finished updating.

Refer to this section in the IP to ensure that the refresh tasks avoid issues with deadlocking and timeouts.

Additional context