Remove model comlexity

we decided to use a constant settling resistance R instead of varying it with overburden and/or density of a snow layer.

Tests still fail, need to be adapted to accept new keyword arguments.