Fix voigt DomainError during optimization

[garrekstemo]

Summary

Use abs on σ and γ width parameters inside the voigt function to prevent DomainError when the optimizer explores negative parameter space.

Problem

The Thompson et al. (1987) FWHM approximation uses odd powers (fG^5, fL^5) that preserve the sign of the input. When σ or γ go negative during optimization, f5 becomes negative, and f5^(1/5) raises a DomainError because Julia can't raise negative reals to fractional powers.

Other models (gaussian, lorentzian) don't have this issue because they square their width parameters (σ^2, Γ^2), making them always positive.

Test plan

• All 262 CurveFitModels tests pass
• Fixes QPSLab CI issue #8 (voigt model fit returns 500)

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (ee8dcc1) to head (a1eabd3).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##              main        #5   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            2         2           
  Lines          101       101           
=========================================
  Hits           101       101           

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.