Test Fe_bct_LO should be made "cleaner"
The test runs the inp file twice and copies cdn1 to rhomat_inp in between. Both runs are to perform 4 iterations (according to the inp file). But the behavior of the runs has changed due to the changed meaning of the strho switch. Before the meaning of the switch was changed the first run only created the starting density.
The problem is now the following: If 4 normal iterations are performed the Broyden files are generated. But these files have to be deleted before the noco mode is used. Otherwise you get an error. The test works anyway because after the starting density is generated the nocomode is already activated in the first run. So, Fleur tries to load the rhomat_inp file which does not exist. An error in loading this file is the result and thus the run ends after generating the starting density. In the second run now everything is okay because only the starting density was generated.
So the test works because errors are thrown. This will probably break if at some point someone introduces a new feature in fleur to make the program more stable, e.g., a single file for the density instead of cdn1 and rhomat_inp. That would be hard to "debug".