chore: add entity class examples to detection and augment prompts

[lipikaramaswamy]

Consolidate entity label examples into a single ENTITY_LABEL_EXAMPLES dict in engine/constants.py, used by validation, augment, and replacement prompts. Fixes conflicting augment prompt wording and aligns prompt example outputs with their Pydantic schemas.

Summary

• Add entity class examples to validation and augment prompts so the LLM knows what each label means (e.g., city: Houston, San Diego, Doha, Lahore), improving reclassification accuracy
• Consolidate the duplicate _EXAMPLE_LOOKUP (replace) and DEFAULT_ENTITY_LABELS (detection) into a single ENTITY_LABEL_EXAMPLES dict in engine/constants.py
• Fix conflicting augment prompt wording: labels are now "preferred" rather than "the only classes that can be proposed," preserving the ability to create new snake_case labels
• Align example outputs in validation and augment prompts with their Pydantic schemas ({"decisions": [...]} and {"entities": [...]} instead of bare lists)

Type of Change

• Bug fix
• New feature
• Breaking change
• Documentation update
• Refactoring

Testing

• Tests pass locally
• Added/updated tests for changes

Related Issues

Closes #10