Regarding our submodularisation efforts, HydPy-GA-GARTO is an early prototype, so it doesn't reflect the final state of the now-evolved submodel concept. We currently have separate versions that only work as a stand-alone model (ga_garto) or a submodel (ga_garto_submodel1), while later application models like evap_fao56 can be used in both ways. One downside is that we would have to duplicate all integration tests. However, we only test ga_garto thoroughly, so modifying the models is risky.

I think we should merge ga_garto and ga_garto_submodel1 into one application model that works both as a stand-alone model and a submodel to improve consistency and maintainability.