Line color in model graph is wrong when a variable is transformed

[jobrachem]

Take the example from this tutorial: https://docs.liesel-project.org/en/latest/tutorials/md/01a-transform.html

The graph after transformation looks like this:

• The black edges pointing from a and b to sigma are correct, because the model accounts for the possibility that these parameters could be used in the bijector.
• The black edges pointing from a and b to sigma_transformed are wrong. They should be grey, because these parameters are parameters of the distribution.

It is currently not clear whether this is an error in the plotting functionality or in the underlying model functionality.

[hriebl]

Maybe this problem is best solved by splitting Var.all_input_vars() into Var.all_value_input_vars() and Var.all_dist_input_vars(). Then this check can be reduced to if edge[0] in edge[1].all_dist_input_vars(): ... etc. Otherwise too much complex logic is needed in the plotting functions to find out where a Var really points...