MBTA training set for HasSSM needs to be updated

Ines pointed out that the SSM injection-recovery was affected by a bug!

Full message from mattermost: While having a deeper look into the HasSSM injections analysis, we noticed that there was a bug in the offline configuration used to provide the MBTA results on these injections... This unfortunately deeply affects the statistics and results of the whole HasSSM chunk processing. We reprocessed it entirely after fixing the bug, and you can find the .xml files for the correct injections at Caltech in /home/ines.bentara/SSM/HasSSM_injections/coinc_02/.

This will require a change in the dataset and consequently retraining of the classifier.