Refactor boundary flux/value methods (again)
To hopefully make it a bit more clear and also re-add the boundary layer as a dispatch argument for boundaryvalue
, since some implementations (e.g. SEB) might only make sense for one of the upper or lower boundary.