Update photom schema

[tddesjardins]

In the photom schema, we have under phot_table an entry for each optical element that currently contains the following:

• photmjsr
• uncertainty
• pixelareasr

I suggest we add to this the following fields (welcome feedback here):

• wavelength
  • title: Throughput Wavelength Array (micron)
  • description: Array of wavelengths in units of microns at which the throughput has been measured.
  • dtype: array
  • items:
    • dtype: float
• throughput
  • title: Throughput Array (unitless)
  • description: Array of unitless throughput measurements. The throughput is the integrated throughput including all elements of the optical chain from the entrance aperture to the detector pixels including reflectivity, bandpass throughput, and detector quantum efficiency. For the spectral elements GRISM and PRISM, this corresponds to the first-order of the dispersed spectrum.
  • dtype: array
  • items:
    • dtype: float

One thing I have not figured out yet is how to store the additional zeroth-order throughput of the grism which we additionally have. Any suggestions for that are welcome. Maybe an optional field throughput_zero ?

[schlafly]

The Goddard throughput files are in effective area, which I find kind of nice. This is degenerate at some level with photmjsr, but all things being equal, I'd rather have the normalization in the throughput as well. Is there a reason not to include that?

Making sure I understand---the zeroth order grism is a ~direct image with weird-ish but fairly nominal PSF? Do higher grism orders come in as well at some very low amplitude?

At some level this is vaguely analogous to ghosts in direct imaging, which I feel like we usually do not include in the PSF or throughput curves, arguing that they are uncertain and small. In that context, one option is to not include that information. But I gather that the amplitudes are larger for the grism zeroth order data and we should include it.

I don't have a good idea there. I have a vague preference for not complicating the schema, and instead adding a new "filter" that is "grism_order_zero" that gets all of the information. It's unclear what we would do with that information in romancal, but at least the data would be present in the reference files.