fix formatting of unit after appending angle

[cshanahan1]

This PR fixes a traceback when changing flux units via API. When changing between Jy and erg/c/cm2/A, the traceback said the unit was not in the list of available options for the y axis because the strings differed. This happens when handing contours. To reproduce:

from jdaviz.conftest import _create_spectrum1d_cube_with_fluxunit
from jdaviz import Cubeviz
import astropy.units as u
cube = _create_spectrum1d_cube_with_fluxunit(shape=(25, 25, 4), fluxunit=u.Jy / u.sr, with_uncerts=True)
cubeviz_helper = Cubeviz()
cubeviz_helper.load_data(cube, data_label="test")
uc = cubeviz_helper.plugins['Unit Conversion']._obj
uc.flux_unit.selected='erg / (Angstrom s cm2)'

In the unit conversion plugin, the '_append_angle_correctly' method to combine chosen flux and angle units sometimes produces unit strings that aren't in the required order for setting the spectral axis. These options are hard coded in UnitConverterWithSpectral (app.py). Additionally, this method was setting the sb_unit_selected read-only attribute twice in a row, so this redundancy was removed.

There is a remaining traceback coming from contours when changing flux unit, which is described in JDAT-4785.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 89.60%. Comparing base (d49e3b5) to head (e824d32).
Report is 3 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3198      +/-   ##
==========================================
+ Coverage   88.50%   89.60%   +1.09%     
==========================================
  Files         124      124              
  Lines       18448    20869    +2421     
==========================================
+ Hits        16327    18699    +2372     
- Misses       2121     2170      +49     

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

[pllim]

Maybe add your reproducing code to regression test suite?

[kecnry]

I think this will also be fixed by #3192 (and if it doesn't, we might want to apply the fix on top of that since it will likely require a different solution)

[cshanahan1]

closing this, it's fixed by 3192