Make the DF-level trigger matching tool be more forgiving about retrieval errors.
There are some very rare events on data whose trigger navigation is a little odd. I've been unable to figure out why they are like this but they really are very few. We can safely ignore them by just failing the trigger matching on these events. However they will right now kill the event with no way of saving it. This MR makes the matching tool more forgiving by default, with the option to re-enable the old behaviour.
Removing WIP. Testing on one of the data files that had problems before this code now just produces a WARNING rather than crashing the job.
edit:
Adding urgent flag as this is a blocker for requesting new derivations.
Edited by Jon Burr