KCWI scattered light

[rcooke-ast]

This PR includes the functionality to generate a (spectrograph dependent) scattered light model, and subtract this light from the input image. The scattered light should be removed before correcting for spatial and spectral illumination, so it has to be included in the rawimage class. Here is an example for the KCWI correction. This is based on the left, right, and central parts of the detector (i.e. the not illuminated parts of the detector). This is a pixelflat.

[codecov-commenter]

Codecov Report

Merging #1661 (9c27c69) into develop (b4f86a5) will increase coverage by 0.06%.
The diff coverage is 23.91%.

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

@@             Coverage Diff             @@
##           develop    #1661      +/-   ##
===========================================
+ Coverage    41.01%   41.07%   +0.06%     
===========================================
  Files          189      189              
  Lines        43471    43583     +112     
===========================================
+ Hits         17829    17903      +74     
- Misses       25642    25680      +38     

Files	Coverage Δ
pypeit/bspline/bspline.py	80.00% <ø> (ø)
pypeit/core/parse.py	55.35% <ø> (ø)
pypeit/par/pypeitpar.py	96.07% <100.00%> (+<0.01%)	⬆️
pypeit/core/flat.py	50.30% <0.00%> (ø)
pypeit/spectrographs/spectrograph.py	67.31% <33.33%> (+0.71%)	⬆️
pypeit/find_objects.py	45.95% <0.00%> (-0.49%)	⬇️
pypeit/core/datacube.py	5.58% <0.00%> (ø)
pypeit/images/rawimage.py	57.18% <16.66%> (-1.32%)	⬇️
pypeit/spectrographs/keck_kcwi.py	29.81% <22.22%> (-0.38%)	⬇️

... and 23 files with indirect coverage changes

[kbwestfall]

A few minor comments from me, and two questions:

• Can you add a description of this approach to the keck_kcwi.rst doc?
• Can we save the scattered light image somewhere? In either or both of the Flat calilbration frame and the spec2d files?

[jhennawi]

Very minor comments. Thanks!

[rcooke-ast]

Thanks @jhennawi and @kbwestfall - I've addressed your comments.

In the process, I have also merged in in PR #1678 to take advantage of the new way to document changes.

[rcooke-ast]

Sorry @kbwestfall, I forgot to address your other questions...

• Can you add a description of this approach to the keck_kcwi.rst doc?

Yes, I will add this to Keck/KCWI docs.

• Can we save the scattered light image somewhere? In either or both of the Flat calilbration frame and the spec2d files?

I think this is not a good idea for the moment. I am working on a parameterised model that would need to store the model parameters, rather than an image (together with a routine that can generate the image from the model parameters). So, I'm tempted to wait until the next round of KCWI scattered light removal to incorporate this.

[kbwestfall]

On saving the scattered light image, I'm fine with putting this off, but I wanted to advocate for saving the full image. In similar situations I also used to be of the opinion that it's better to save the model parameters and provide a script if people want to reconstruct the model. But I've come around to the opinion that this presents a barrier (albeit small) to people being able to assess the results. Disk space is cheap and, IMO, enabling people to quickly open the file with astropy and view it with matplotlib is worth the cost of a larger output file.

[kbwestfall]

Approving!

[rcooke-ast]

Tests pass

[jhennawi]

Approved. Thanks!