Remove duplicated local census records on deployment

[Senen]

References

Related Pull Requests: #3646

PR #3646 includes a migration that adds an unique database constraint to LocalCensusRecord model so we need to provide a way to remove duplicated records before to apply the new index to database.

Objectives

Remove duplicated local census records automatically during capistrano deployment and before applying new aforementioned migration.

Also provide a rake task to remove duplicated records manually for those Consul installations that are not using caspistrano.

Release notes ⚠️

With this PR people using capistrano may not worry about deleting local census records dupicates manually from the database, but people using other deployment strategies that capistrano, for example Heroku, should manually run the rake task local_census_records:remove_duplicates before running db:migrate.

Production server test results

Deployment test steps:

1. Prepare a production server with the Ansible installer and deploy the Consul 1.0.0 version.
2. Create dups in the LocalCensusRecord database table
3. Deploy with capistrano the code of this PR
4. Check during deployment and before the capistrano step deploy:migrate those duplicated records were removed successfully

Below the deployment log that demonstrates this PR is working successfully.

01:02 remove_local_census_records_duplicates
      01 /home/deploy/consul/rvm1scripts/rvm-auto.sh . bundle exec rake local_census_records:remove_duplicates
      01 INFO Removing local census records duplicates
      01
      01 INFO Removed 6 records.
      01
    ✔ 01 [email protected] 4.907s
01:07 deploy:migrate
      [deploy:migrate] Run `rake db:migrate`
01:07 deploy:migrating

[javierm]

@Senen Thanks a lot! I've left a couple of comments about minor things.

[javierm]

@Senen Thanks a lot!