Investigate moment map commented test

[pllim]

This PR addresses the following unreleased code (hence no change log needed):

• Removing the need to create temporary spectrum just to convert unit for moment map, hence removing the need to "Centralize pixel scale factor information".
  • Update: The whole changes from Jesse in moment map are now reverted. Moment map will stay in surface brightness.
• Fixing compatibility with unreleased specutils (hence fixing devdeps).
• Fixing a possible bug in flux_conversion function where equivalencies math is flipped for cases len(values) != 2 and len(values) == 2. This is accomplished by having both logic go through the same math (there is really no good reason to have separate math for those cases anyway).

Out of scope:

• Fallback value for PIXAR_SR when it is not found. Personally, I think any fallback is dangerous unless you can be sure you are able to look up the actual value from somewhere (e.g., handbooks, databases, websites) for any given telescope/instrument combo. Otherwise, better to force user input or just disallow it from happening.

xref #2910 #2873

🐱

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 88.78%. Comparing base (e65237f) to head (2a120e2).
Report is 161 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2936      +/-   ##
==========================================
- Coverage   88.79%   88.78%   -0.02%     
==========================================
  Files         111      111              
  Lines       17236    17233       -3     
==========================================
- Hits        15305    15300       -5     
- Misses       1931     1933       +2     

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

[javerbukh]

Do moment maps work for you in this PR? I see an error with all default values in the CubevizExample notebook
UnitConversionError: 'MJy / sr' (surface brightness) and 'MJy' (spectral flux density) are not convertible

[pllim]

Do moment maps work for you in this PR?

Is this using that thingy that is not enabled by default?

[javerbukh]

No, but the auto-collapsed spectrum (using the sum function) is in flux units which means the default units are in flux, while moment map is using surface brightness units.

[pllim]

Didn't @kecnry ask us to just leave it broken until Gilbert's thingy is in or something?

[kecnry]

It probably makes most sense to merge the changes to display units first, rebase this on top of that, and then ensure that it correctly handles SB<>SB conversions correctly without unnecessary code for SB<>flux which won't be supported.

[pllim]

Why is the CI green? What are we not testing?

[pllim]

Re: #2936 (comment)

Okay, so we need to keep this code for now, yes? In that case, it can still be simplified and there is still a bug to be fixed.

[pllim]

I cleaned it up (again). What about now?

[kecnry]

We will be keeping SB<>SB unit conversion but will no longer need SB<>flux (the comments seem to suggest the opposite).

[pllim]

Comments aside (which I can update), the code itself looks okay to proceed?

[pllim]

Lots of duplicated logic in flux_conversion. It was hurting my head, so I ended up just re-writing it to be simpler.

[pllim]

OK I think I am done here until I hear more.

[pllim]

The code here is now part of #2940