Suggestion: Add constraint checks on agent movement

[Ozaq]

This is a suggestion that has reached us via Email (paraphrased):

Choosing a larger delta T for the simulation might result in an error message from the Simulation about Agents being outside of the accessible area of the simulation. It is suggested to constraint movement in such a way that updates that would let the agent leave the accessible area are shortened so that agents remain inside of the accessible area.

[Ozaq]

This is something we discussed internally and we had the same idea. In the end we choose not to constrain the movement update for two reasons:

1.) Some pedestrian models are explicitly declared collision free in the literature hence we consider collision handling something in the scope of the model. And aas you have noticed the model starts to lose this property once you start increasing the time step.

2.) We want to provide unambiguous results, meaning we either can correctly update the iteration or we error out.

The defaults provided are expected to work (if they don't then this is a bug) otherwise we want to clearly express breakdown of guarantees.

At a later point in time we might want to rethink this.