Provide a mechanism for granting usage to schemas in on-run-end

[drewbanin]

Feature

Feature description

If a user makes use of custom schemas, then on-run-end cannot currently be used to grant usage to the schemas on Redshift/Snowflake/Postgres. This problem arises because there are N schemas, but the only schema available in the context is the schema defined in the target.

dbt should provide in it's context a list of schemas that were built into during the dbt run.

Example usage:

on-run-end:  {% for schema in schemas %}grant usage on {{ schema }} to PUBLIC;{% endfor %}

We probably will want to namespace schemas, but this is the general idea.

Who will this benefit?

This will be useful specifically for Redshift users. Currently, granting usage to a schema must happen in post-hooks, and Redshift appears to raise concurrent transaction errors if usage is granted to a schema while other models are being created/dropped in that schema (ie. with threads > 1)

[beckjake]

Fixed in #1028