Don't force configuration_out

Right now, file handling is much less robust then before the recent changes. Most issues arise from the fact that the configuration_out node is expected in both the raw data- as well as the interpreted data file. This makes a scan following an aborted or crashed scan crash immideately, since it tries to load the configuration from the raw data file of the previous scan, which does not have the configuration_out node..

also, raw data files from crashed scans cannot be analyzed, solely because the configuration_out node does not exist. These cases should be handled, maybe give out a warning and copy configuration_in to configuration_out if necessary. Again, having a source scan fail after 50 minutes and not being able to interpret the data even though the raw data itself is fine is extremely frustrating...