I stripped out preCICE from the solver and enforce the flux values of the monolithic solver as boundary condition (see directory appl/coupling-ff-pm/iterative-reversed-mono-flux-no-precice).
I assume/hope that I have extracted the monolithic flux values correctly.
I added some code to compare the VTU files in postprocessing. The corresponding CMakeFiles.txt is not in Dune/Dumux style. The code is compiled every time the project is configured.