Remove zeros=nans from objectModel3D in C++ and HDevelop samples

[SatjaSivcev]

Our Genicam producer sets the invalid points to (0,0,0). This means the output object model 3D has many points in (0,0,0), which need to be removed by the user/programmer. For example, in the case of the surface-based matching, the algorithm will slow down and the 3D edge extraction will fail. That means, if we do not remove the invalid points (or the zero's), the user/programmer has to do it by hand (for example using the operator select_points_object_model_3d()). But, this is often forgotten by the user and therefore worsens the performance.

Before we can remove zeros from the SDK we should remove them here.

[chrisasc]

I've been talking to MVTec about this and they said that several customers of them forget to remove these zero points. This makes later processing slow and/or wrong. I think it is a good idea to implement this