One of the requests from the PFSA01894911_20200619 comments was to add Subaru-specific spectroscopy cards:

- Spectroscopy keywords
  All spec. keywords are missing in the present header
  While most of them are for slit spectroscopy and may be omitted for PFS,
  (->needs to update Subaru rule. what other spec instruments do?)
  necessary information such as wavelength range and orientation of dispersion
  should be covered by some basic or instrument-dependent keywords

  WAVELEN
  WAV-MAX
  WAV-MIN
  SLIT
  SLTCPIX1
  SLTCPIX2
  SLT-LEN
  SLT-PA
  SLT-WID
  DISPAXIS
  DISPERSR
 

DISPAXIS is easy and I'll do it right away: 1 for CCDs, 2 for NIR. Note that after DRP IRP the NIR exposures will probably be rotated to have dispersion along columns, to match the CCDs. So the DISPAXIS card would only be valid for raw data.

I take it we can ignore all the SLit cards. For now, at least.

For DISPRSR and the three WAV cards can we use names and values from the first measurement paper, which also show up in Yabe-san's curated tables in spt_exposureTimeCalculator? Those are for the full bandpass (not between the 50% levels of the dichroics, etc.), and will not be good to better than several nm:

In particular, for the WAVELEN card can we just use the central wavelength of the bandpass, or will we need to figure out which wavelength falls on the central row/column? The NAOJ examples have values down to 0.001 nm, which worries me 

We need some convention for DISPRSR. A proposal, based on the NAOJ example would be "VPH_$fringesPerMm_$centralWave"? Add "low" vs "high" for red? I.e "VPH_711_519nm" or "VPH_low_557_805nm"? The red medium assembly is slightly different (adds prisms) but does it matter?