Reopen and post note on incidents for which automatic pipeline retry failed
Spin-off from #217 (comment 1393639550).
When a pipeline fails, check if an incident already exists: that probably means the retry failed => in that case we could reopen the incident and post a comment to state that the automatic retry failed and it needs human investigation (that would also solve the infinite retry loop that we had when the registry was failing).
Edited by Rémy Coutable