When I said to check it by "traditional" slope methods - refer to oil tank analysis in Lambe and Whitman - then put in the hard layer and your "programme" would give a result - could consider to use wedge analysis if you wish - all predicated that the hard layer is "hard!!".
If the distance to the hard layer will not permit the full failure surface to develop, then to use traditional analysis is not correct.
Can also look up Section 4.8 of Fang's Foundation Engineering Handbook, 2nd edition. This gives some insight to the problem although they do not reference one of the original papers (by Button, 1957) on the issue. Hope this gives you some insight into the problem. But a thin layer, as I suggested might be better considered to be squeezing.