`imProc.image.image2subimages` and `AstroImage.crop` should attempt to update WCS

[EastEriq]

even approximately, taking into account at least the center offset and not the rotation/distorsions, which may be small. Currently they don't, and therefore functions like imProc.astrometry.astrometryRefine on image subsets fail

[EastEriq]

And perhaps also augment the header with CROPID, see issue #210

[EastEriq]

Example of what I'm forced to do, and should be automatic and foolproofed:

    % split in subimages
    AI=imProc.image.image2subimages(AI,2048);
    % correct CPIX for the second image (temporary workaround for
    %  https://github.com/EranOfek/AstroPack/issues/182 )
    AI(2).WCS.CRPIX(2)=AI(2).WCS.CRPIX(2)-2037;
    AI(2).HeaderData.setVal('CRPIX2',AI(2).WCS.CRPIX(2));
    AI(1).HeaderData.setVal('CROPID',1);
    AI(2).HeaderData.setVal('CROPID',2);

[EranOfek]

AstroImage.crop & imProc.image.image2subimages - feature added (exact) Need to test!

[EranOfek]

crop tested

[EastEriq]

apparently commit https://github.com/EranOfek/AstroPack/commit/f0a4ec1b1c7e064f5a24bd7611fea55ae37621f2 created a regression. Reopening and reporting my investigations here

[EastEriq]

As a baseline, this call is ok:

AI=AstroImage('/raid/enrico/PTF/2ndTake/images/level1/proc/2012/01/07/f11/c0/p5/v1/PTF_201201075069_i_p_scie_t120955_u011063047_f11_p015199_c00.fits');
AI=imProc.image.image2subimages(AI,2048);
for k=1:2
    ds9(AI(k),k,'zoom','to fit')
end

In the sense that looking in ds9, the coordinates of the upper two corners of the first frame match those of the lower two corners of the second. (I don't know if there wouldn't be a function to check it instead of hovering the mouse) In a more complex script of mine, not. Going to figure out why.

[EastEriq]

ok, "not your fault". The problem funnily pops up when reading the original image mask (3rd line of the following):

ff='/raid/enrico/PTF/2ndTake/images/level1/proc/2012/01/07/f11/c0/p5/v1/PTF_201201075069_i_p_scie_t120955_u011063047_f11_p015199_c00.fits';
AI=AstroImage(ff);
AI.MaskData.Data=mapPTF2oursMask(MaskImage(strrep(ff,'scie','mask')));
AI=imProc.image.image2subimages(AI,2048);
for k=1:2
    ds9(AI(k),k,'zoom','to fit')
end

Weirdly even if the mask refers to the full image, only the coordinates of the second subimage come out wrong.

Reclosing this issue and opening a dedicated one.

[EastEriq]

crop tested

Not tested well enough, apparently.