`/retest` should only re-run failed pipelines

[pierDipi]

If a few pipelines are failing, there is no way for us to re-test only the failed ones.

Restarting all of them is a waste of resources and increases the chance of getting transient errors on pipelines (registry temporarily down, etc)

This would be similar to how Prow works: https://github.com/kubernetes/community/blob/master/contributors/guide/pull-requests.md

[chmouel]

That's a good point but unfortunately this ship has sailed since we can't change the default behavior, we can tho introduce a /retest failed (assuming there is no PR named failed)