Fix PossibleClassicalForms for nonzero traces

[fingolfin]

When both trace(g) and trace(g^-1) were nonzero the function returned early without testing the coefficient identities, so it never ruled out a form. Over a large field almost every element has nonzero trace, which left recog's classical recognition of SL(6,257) undecided about half the time (see gap-packages/recog#492).

Nonzero traces determine the scalar lambda relating g^-1 to g directly, so use it in the existing coefficient test instead of returning.

Co-Authored-By: Claude Fable 5.1 noreply@anthropic.com

[codecov]

Codecov Report

❌ Patch coverage is 77.77778% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 81.83%. Comparing base (4532e07) to head (e5d33ee).

Files with missing lines	Patch %	Lines
lib/recognition.gi	77.77%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master      #88      +/-   ##
==========================================
+ Coverage   80.40%   81.83%   +1.42%     
==========================================
  Files           8        8              
  Lines        3670     3671       +1     
==========================================
+ Hits         2951     3004      +53     
+ Misses        719      667      -52     

Files with missing lines	Coverage Δ
lib/recognition.gi	56.43% <77.77%> (+11.27%)	⬆️

🚀 New features to boost your workflow:

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