Add a separate failure reason for image pull failures

Description

One of our customers (internal) reported that failed pull images are marked as script_failure. However, based on our descriptions here, script_failure should only be used if the script defined in the .gitlab-ci.yml file failed.

This makes it difficult for them to use the retry keyword when retrying the job due to a failed image pull, as this will also retry their job if the script failed.

Steps to reproduce

I created a test project here: https://gitlab.com/jdasmarinas/zd-326066/-/jobs/3045919061

test:
  image: nonexistent/docker-image:non-existent
  script:
    - ls

Then if you look at the job API, you will notice that failure_reason is set to script_failure.

Proposal

We should have a dedicated failure reason for image pull failures. As of right now, we don't have a reliable way to identify image pull failures from actual script failure, which makes it tricky to only retry the job if there was an issue pulling the image.