The current implementation of makeSkyWcsFrameDict takes the Mapping out of the pixelsToFieldAngle Transform and inserts it into the resulting SkyWcs which loses the Frames that were in that Transform. If we create a SkyWcs from the boresight and camera geometry, we want to maintain all of the Frames and their Mappings, so that we can individually access them later.
I'm not sure what the best approach for this is, given the options for constructing the AST FrameDict. It might be a simple change. Russell Owen: can you please give some input?