Add link requirement and unknown category isolation

MR adds two new validation checks to the features.yml file:

1. Unknown category isolation: ensure that the unknown category is not used with other categories.
2. Require a link for all features unless category is unknown. Also fixes 19 violations.

Related to Product#14293 (closed)

Duo summary

Features Checker (lib/features/features_yml_checker.rb)

New Functionality Added:

1. Unknown Category Isolation Validation

   • Added check_unknown_category_isolation method
   • Prevents mixing "unknown" with other categories
   • Error message: "category 'unknown' cannot be combined with other categories: [list]"

2. Stricter Link Requirement Logic

   • Updated category_is_unknown? method to only return true when category is exactly ['unknown']
   • Previously: returned true if "unknown" was anywhere in the categories array
   • Now: returns true only if the entire categories array is ['unknown']

Performance Optimizations:

3. Single Category Extraction

   • Added extract_categories method using Ruby's Array() pattern (consistent with existing codebase)
   • Categories now extracted once per feature instead of 3 times
   • All validation methods updated to accept categories as parameter

4. Method Signature Updates:

   • check_category_validity - now takes categories parameter
   • check_link_requirement - now takes categories parameter
   • check_unknown_category_isolation - now takes categories parameter
   • category_is_unknown? - now takes categories parameter instead of feature_data

Integration:

5. Updated Validation Pipeline
   • Added check_unknown_category_isolation to the validation sequence
   • All methods now receive pre-extracted categories

Test File (spec/features/features_yml_checker_spec.rb)

Test Coverage Improvements:

1. Removed Redundant Tests (earlier in conversation):

   • Removed duplicate positive validation tests
   • Removed redundant link validation tests
   • Streamlined from ~15 tests to 11 focused tests

2. Added New Test Cases:

   • 'fails when unknown is mixed with valid categories' - expects 2 errors (isolation + link requirement)
   • 'fails when unknown is mixed with invalid categories' - expects 3 errors (invalid category + isolation + link requirement)
   • 'passes when multiple valid categories are used' - validates array categories work correctly
   • Kept existing 'passes when link is missing but category is unknown' test

Test Expectations Updated:

3. Accurate Error Counts
   • Tests now expect correct number of validation errors
   • Account for cascading effects of stricter category_is_unknown? logic

Real Data Impact (data/features.yml)

Validation Results:

• Before changes: 19 validation errors (missing link fields)
• After changes: 0 validation errors ✅
• Errors fixed: 19 errors resolved

The real data/features.yml file now passes validation completely, indicating:

1. No features currently mix "unknown" with other categories
2. All features that were missing links have been fixed
3. The new validation rules don't break existing valid data

Overall Impact:

• ✅ Functionality: Enhanced validation with semantic correctness for "unknown" categories
• ✅ Performance: ~67% reduction in category extraction operations
• ✅ Quality: Comprehensive test coverage with no redundancy
• ✅ Compatibility: All existing valid data continues to pass validation
• ✅ Standards: Code follows project conventions and passes rubocop

The implementation successfully adds new validation rules while maintaining backward compatibility and improving performance.