Add support for Nonlinear loops

[kabdelhak]

The first goal is a simple model such as

model nonlinearLoop
  Real x,y;
equation
  x = y^2 + sin(time);
  y = sin(x);
end nonlinearLoop;

The basic structure for this already exists. We have jacobians and we have the structures for StrongComponent loops. Currently we only have residual variables for DAE-Mode, but i think there is no reason to not use the deaResidual of an equation as general residual.

Maybe we should even create a residual variable for each equation to have a unique identifier for them. Could be benefitial.