Change warnings and errors related to steady_state_model
Even substituting the analytical expression for k in
k=log(k); in a recursive definition (see #554 (closed)) does not help, because it leads to
ERROR: in the 'steady_state' block, variable 'k' is declared twice
I would prefer to relax this last error and make it a warning. There is no reason why we should disallow overwriting previous computations.
At the same time, we do not provide a warning, if not all variables have been defined within a
steady_state_model. Missing values are simply taken to be 0. This makes debugging of larger models hard. I would suggest to add a warning in this case.