Add spectrum_required=False to FormatXTC

[phyy-nx]

New parameter and updated behavior to account for occasional dropped spectra in the XTC stream, when spectra a used to produce a calibrated average wavelength. If spectrum_required=True, and the FEE spectrometer is not found in an event, get_spectrum will raise a RuntimeError. If spectrum_required = False and spectra calibration constants (spectrum_eV_per_pixel and spectrum_eV_offset) are provided, wavelength_offset can be used to apply a general correction for any events with a dropped spectrum. If the spectrum is present and calibration constants are provided, wavelength_offset is ignored.

[codecov]

Codecov Report

Merging #674 (1522982) into main (dbd7ede) will decrease coverage by 0.03%.
The diff coverage is 7.14%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #674      +/-   ##
==========================================
- Coverage   42.62%   42.60%   -0.03%     
==========================================
  Files         182      182              
  Lines       15924    15935      +11     
  Branches     3056     3058       +2     
==========================================
+ Hits         6788     6789       +1     
- Misses       8510     8520      +10     
  Partials      626      626              

[dwpaley]

We've used this successfully a few times, so as such it LGTM. A couple comments that don't need to prevent merging:

• This changes the handling of wavelength_offset in the case where the spectrum is present and calibration constants are provided. (namely it was applied before, and now it isn't.) Is there any existing use case for wavelength_offset where this change will break backwards compatibility? (I can't think of any)
• It's starting to accumulate a maze of nested try/excepts in there, maybe this is a candidate for some mild refactoring. I'm sure we will have more time in a couple months :)

[phyy-nx]

@dwpaley

This changes the handling of wavelength_offset in the case where the spectrum is present and calibration constants are provided. (namely it was applied before, and now it isn't.) Is there any existing use case for wavelength_offset where this change will break backwards compatibility? (I can't think of any)

No, we've never had a situation with both a calibrated spectrum and the need to apply wavelength_offset. If we did, we'd just add wavelength_offset to spectrum_eV_offset (adjusting the units).