lvmSFrame sky and sky_error extensions

The current quick_sky_subtraction routine is saving the SkyE flux and error as the sky and sky_error extensions for the lvmSFrame. These should be the sky that was subtracted from each fiber in the lvmCFrame (skysub) and associated error (np.sqrt(error^2+skysub_error^2)).

In master branch, the current code is (note the TODOs):

# TODO - deal with error in sky-subtracted data ; replaced "error_c"
# TODO - deal properly with the error, sky, and sky_error
# this is incorrect
error_c = cframe._error
sky_c = cframe._sky_east
sky_error = cframe._sky_east_error

print("writing lvmSFrame")
sframe = lvmSFrame(data=skydata, error=error_c, mask=cframe._mask, sky=sky_c, sky_error=sky_error,
                   wave=cframe._wave, lsf=cframe._lsf, header=cframe._header, slitmap=cframe._slitmap)

I think we have several issues to sort out with regard to the code here. Here is more of the region that Amy cited in SkyMethod.

 # create spectrum for sky subtraction
    scisky = create_skysub_spectrum(sky_hdu, tel='sci', method=skymethod)
    skyesky = create_skysub_spectrum(sky_hdu, tel="skye", method=skymethod)
    skywsky = create_skysub_spectrum(sky_hdu, tel="skyw", method=skymethod)
   # select correct fibers
    data = cframe._data
    slitmap = Table(cframe._slitmap)
    scifibers = slitmap[slitmap["telescope"] == "Sci"]["fiberid"] - 1
    skyefibers = slitmap[slitmap["telescope"] == "SkyE"]["fiberid"] - 1
    skywfibers = slitmap[slitmap["telescope"] == "SkyW"]["fiberid"] - 1

    # sky subtract each spectrum
    skydata = data.copy()
    skydata[scifibers] = data[scifibers] - scisky
    skydata[skyefibers] = data[skyefibers] - skyesky
    skydata[skywfibers] = data[skywfibers] - skywsky

    # write out sky table to ancillary file
    mjd = cframe._header['MJD']
    expnum = cframe._header['EXPOSURE']
    tileid = cframe._header['TILE_ID']
    skytable = path.full('lvm_anc', mjd=mjd, tileid=tileid, drpver=drpver,
                         kind='sky', camera='brz', imagetype='table', expnum=expnum)

    sky_hdu.writeto(skytable, overwrite=True)

    # TODO - deal with error in sky-subtracted data ; replaced "error_c"
    # TODO - deal properly with the error, sky, and sky_error
    # this is incorrect
    error_c = cframe._error
    sky_c = cframe._sky_east
    sky_error = cframe._sky_east_error

    print("writing lvmSFrame")
    sframe = lvmSFrame(data=skydata, error=error_c, mask=cframe._mask, sky=sky_c, sky_error=sky_error,
                       wave=cframe._wave, lsf=cframe._lsf, header=cframe._header, slitmap=cframe._slitmap)
    sframe.writeFitsData(out_sframe)
    # TODO: check on expnum=7632 for halpha emission in sky fibers

It seems to me what we are using for creating the sky for the telescopes is what we want. Either we should use the sky for the science telescope or we should have a procedure.
We need to decide what we want in the sky extension in the SFrame. I would argue it should be whatever we subtracted from the data in the CFrame, that is sky_sci, skyesky, skywsky for the appropriate fibers.

sdss / lvmdrp

lvmSFrame sky and sky_error extensions #81