lc.to_fits() + lk.io.read() does not preserve correct FLUX_ORIGIN meta data for SAP_FLUX data

[pmaxted]

If SAP_FLUX data are downloaded and saved as a FITS file then re-loading the data using lk.io.read() assigns the incorrect value to lc.meta['FLUX_ORIGIN']

Example

>>> import lightkurve as lk
>>> s = lk.search_lightcurve('HD1',author='TESS-SPOC',exptime=200)
>>> lc = s.download_all(flux_column='sap_flux').stitch()
>>> print(lc.meta['FLUX_ORIGIN'])
>>> print(lc.flux[42])
>>> lc.to_fits('test.fits',overwrite=True)
>>> lctest = lk.io.read('test.fits')
>>> print(lctest.meta['FLUX_ORIGIN'])
>>> print(lctest.flux[42])

sap_flux
0.9995154142379761
pdcsap_flux
0.9995154142379761 electron / s

Expected behavior

lc.to_fits() needs to write the FLUX_ORIGIN value to the header of the output FITS file so that lk.io.read() can read this keyword and assign the correct value to this item in the meta data.

Environment

• platform: Windows
• lightkurve version: '2.4.1'
• installation method: pip

[orionlee]

The issue is essentially a symptom of the underlying problem, discussed in #896