afocal systems break focal length and platescale calculations

[douglase]

As discovered recently by @neilzim and as previously noted but not documented, the current way plate scales are calculated in the fresnel wavefront class depends on the focal length attribute carried through the sequence of propagations by the wavefront class. Thus, it breaks when you have an afocal system where plane-to-plane propagation occurs.

We need either another way of carrying focal length through a system, or more graceful failure and better documentation.