Rudder

relayd has a feature that allows to send old report it received while it was shut down. But that feature lead to complicated problem reguarding compliance (#17180).

So until we are able to compute compliance reliably for reports in the past, historize it correctly, and know how to deal with edge cases, we relay should only send old reports in order it got them and never send a report received before an other if that other was already sent. So: when relayd restart, do whatever but in reception chronological order, and on permanent run, only send new runs.