Fix moment calculations

[rosteen]

This solves a bug that was reported offline by @camipacifici - moment 0 should be in units of flux * spectral axis, being the flux integrated over the spectral range, matching the units of line flux. This PR changes from doing a simple sum of flux for moment 0, to summing flux*dx as in line_flux.

The case for SpectrumCollection is because the spectral_axis of a SpectrumCollection comes out as SpectralCoord, and we need bin_edges from SpectralAxis in each individual member of the collection.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 86.83%. Comparing base (28b0cdb) to head (4d6c35f).
Report is 10 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1141   +/-   ##
=======================================
  Coverage   86.82%   86.83%           
=======================================
  Files          63       63           
  Lines        4525     4528    +3     
=======================================
+ Hits         3929     3932    +3     
  Misses        596      596           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[rosteen]

Actually, the higher orders also need a similar correction. The sum should have the terms multiplied by dx in those as well, with the additional spectral unit getting cancelled out by dividing by the (corrected) M0.

[rosteen]

There's an unrelated deprecation warning in devdeps that I'll address separately.

[camipacifici]

It is now correct in the 1D case for flux density and it seems to propagate well in Cubeviz. Thank you!

[rosteen]

There's an unrelated deprecation warning in devdeps that I'll address separately.

Ah, FWIW this is upstream in GWCS: spacetelescope/gwcs#499.

[pllim]

Looks like this broke tests over at Jdaviz, see spacetelescope/jdaviz#2888 (comment)