Metrics for pipeline jobs

Some jobs, especially the ones from the security products like SAST, are fetching docker images for each run. These images are moving in time, and are not fixed for each GitLab release because they contain some fresh security advisories data. This also means some jobs can start failing with no warning, and we don't have any clue of that currently, until a customer or some internal team will raise the issue to us.

The Security Products team would like to have some metrics about successes and failures to start, or even better: some entries directly in Sentry (in a specific group).

/cc @bikebilly