Split the loading of concatenated feature count tables in two steps

[kevinrue]

1. concatenate the tables and write the result to a compressed text file (cluster job)
2. load to database (head node)

[snsansom]

Hi Kevin,

This looks good - as you proposed we should also add the active_if statements so that loading of results to the database can be disabled for large samples.

Best wishes,
Steve

[kevinrue]

The last commit seems to have taken care of it on my side, tested on a project. Perhaps you want to test it yourself on a toy project to double check before merging?