Fix fit_lines with single quantity as window value

[rosteen]

Fixes #1160. Currently on main, giving a single Quantity as input for window gets ignored, since a Quantity is neither an int nor a float. You can see below that one of the fits in the fitting tests is actually very bad, since it's fitting the whole spectrum and doesn't limit to a window around the left gaussian:

With this fix, the fit for the left gaussian in the test case properly gets limited to the defined window and is fit well:

[rosteen]

Looks like the one test failure is a remote data timeout.

@TurboSlayer would you check to see if this fixes your use case?

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 86.89%. Comparing base (cadfe73) to head (6ada51c).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1164   +/-   ##
=======================================
  Coverage   86.88%   86.89%           
=======================================
  Files          63       63           
  Lines        4544     4547    +3     
=======================================
+ Hits         3948     3951    +3     
  Misses        596      596           

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

[TurboSlayer]

Thanks - I managed to work around it by setting manual bounds anyways. I trust from the screenshots that you've fixed the issue haha.

[pllim]

Adding test to fit in frequency space might be good, or a test where spectral_axis is wavelength but the window in frequency (is that even allowed) but you can also argue expanding test suite is out of scope here.

Needs a change log?

[rosteen]

Needs a change log?

You're not my boss!

(yes, yes, I'll add one momentarily)

[pllim]

You're not my boss!

Indeed I am not. But I can unleash the change log bot here if you want. 😉 (#1165)

[pllim]

Remote data failure unrelated and can be ignored.