Fix incorrect license detection in dragon4.c

[Kaushik-Kumar-CEG]

Fixes #4584

Description

This PR fixes an incorrect license detection in NumPy's dragon4.c, where the text "As a special exception" was incorrectly triggering a generic-exception match.

Changes:

• Modified license-intro_8.RULE: Downgraded this rule from is_license_intro to is_license_clue. This ensures the phrase contributes to evidence but does not trigger a false positive when found alone.
• Added false-positive_dragon4.RULE: Added a specific false positive rule to ignore the misleading text ("As a special exception the smallest normal value") found in dragon4.c.

Verification:

• Reproduced the issue locally: confirmed generic-exception was originally detected.
• Verified the fix: confirmed the detection is gone after the changes.
• Validated the rule downgrade: confirmed that license-intro_8.RULE correctly behaves as a clue (no detection) when isolated.

Tasks

• Reviewed contribution guidelines
• PR is descriptively titled 📄 and links the original issue above 🔗
• Tests pass -- look for a green checkbox ✔️ a few minutes after opening your PR (Verified fix locally)
• Commits are in uniquely-named feature branch and has no merge conflicts 📂
• Updated documentation pages (if applicable)
• Updated CHANGELOG.rst (if applicable)

[Kaushik-Kumar-CEG]

Hey @pombredanne @AyanSinhaMahapatra! I looked into the CI failure and ran the license tests locally to verify everything works:

Local test results:

• pytest tests/licensedcode/test_detect.py -v → 55 passed ✅
• pytest tests/licensedcode/ -v → 474 passed ✅

The failing test (test_scan_system_package_end_to_end_installed_win_reg) is for Windows Registry package detection, which is completely unrelated to this license fix for dragon4.c. The CI shows 3595 tests passed with only this 1 unrelated failure.

All license tests pass successfully, so the fix is working correctly without breaking existing functionality. Let me know if you need anything else!