Add index to Task.status

Summary:
We have monitoring and clean-up tasks that engage in table scans because they primarily care about
status; this is currently triggering our long-query warnings and creating noise.
Hopefully an index fixes that.
Size of the table is ~3m items in prod, so the indexing may take a bit, but we don't need
the index immediately, so adding it concurrently seems like a way to avoid downtime.

Test Plan: Staging, potentially a (new?) read replica

Reviewers: naphat

Reviewed By: naphat

Subscribers: changesbot, anupc

Differential Revision: https://tails.corp.dropbox.com/D233244

migrations/versions/2c49f64c0d4a_add_status_index_to_task.py

@@ -0,0 +1,21 @@
+"""Add status index to task table.
+
+Revision ID: 2c49f64c0d4a
+Revises: 43397e521791
+Create Date: 2016-09-29 15:45:51.104377
+
+"""
+
+# revision identifiers, used by Alembic.
+revision = '2c49f64c0d4a'
+down_revision = '43397e521791'
+
+from alembic import op
+
+
+def upgrade():
+    op.create_index('idx_task_status', 'task', ['status'], unique=False, postgresql_concurrently=True)
+
+
+def downgrade():
+    op.drop_index('idx_task_status', table_name='task')