Modelling a pile in slide which improves your safety factor is a fallacy. The first big problem is 2D v 3D. Are you modelling the pile as a soil element and having high (few thousand?) cohesion value?
All this will do is force the failure plan to extend below your pile tip.
In short, limit equilibrium stability programs are not a good way to model retaining walls. Soil nail anchors, etc yes, embedded retaining walls no IMO.