CIDataFit class and updating frame geometry / ci pattern during extract

[James Nightingale]

I am setting it up so that from a CIData, we extract a CIDataFit class, which is the data extracted and used for the fit. The CIData class does not include a mask, which the CIDataFit class does, which allows us to update the mask throughout the pipeline using input mask_functions.

When the CIFit is performed, it is thus perfomed on the extract CIData. This is fine, and works, however there is a problem whereby the frame_geometry and ci_pattern of the CIData's data (e.g. the thing which are CIFrames) are still associated with the values from the CIData. After extraction, the location of various regions of the image (e.g. the parallel overscan, serial overscan, serial trains, the ci_pattern itself) will change.

Thus, the extractioon routines which we associated with a CIFrame / CIData need to be able to generate new CIFrames which upddate the overscan regions, ci_pattern, etc. based on the input extraction. This should be straight forward, given we always know for the parallel / serial extraction whcih direction data is being extracted from.

I worry this may get more complicated, but will have a go at doing this and get a feel for if this is the right solution.