Over the last while the data centre has made ASDF files with two different naming conventions:

• f"{instrument}_L1_{start_time:%Y%m%dT%H%M%S}_{dataset_id}..asdf"
• f"{instrument}_L1_{start_time:%Y%m%dT%H%M%S}_{dataset_id}_user_tools..asdf"

In the future this suffix will be changed to:

• f"{instrument}_L1_{start_time:%Y%m%dT%H%M%S}_{dataset_id}_metadata.asdf"

Therefore we need to adapt the directory loader to handle picking the most applicable of these patterns if more than one is detected in the folder. I think the algorithm should be as follows:

• Glob the directory for all ASDF files
• Group all results by the dataset id in the filename
• Pick the ASDF file to use with the following priority: (_metadata, _user_tools, <nothing>)
• Throw a warning to the user if any ASDF files with older suffixes are found