Define two jobs: One for successful and another one for failed deployments

The following discussion from !1549 should be addressed:

• @nolith started a discussion:

  @mayra-cabrera I'm leaving this note with the intent of converting it into a follow-up item.

  When a deployment job fails, we will send a notification thanks to when: always.

  When we retry the failed deployment job, it will not execute this job again and it will never send the successful deployment message again.

  I think we should do the same thing we are doing in the deployer, where we define 2 jobs, one for successful deployments and one for failed ones (when: on_failure), in this way we have 2 independent jobs and after a failed deployment the successful job can still run on a deployment retry.

  Obviously, we will still have the problem where a second failure will not trigger the message again (but this is exactly the same behavior we have in deployer)