feat: real enforcement for habitability + living-community principles

[oddtitoreal]

Resolves the 4 semantic warnings by adding enforcement, not by silencing them.

State machine (community-space-brand):

• add guard_relational_quality, guard_activation_color_ratio (principle_habitability) and guard_living_community (principle_living_community)
• attach them to real transitions (space_state_selection -> generate_response, generate_response -> response_ready)
• add guard.constraint_ids linking every constraint to an enforcing guard

Model upgrade principle -> constraint -> guard -> transition:

• schema: add guards[].constraint_ids (pattern ^constraint_[a-z0-9_]+$)
• validator: add direct constraint->guard coverage check; add --strict mode (warnings fail) for real, non-fictional examples
• luxury example: map its constraints via constraint_ids to stay warning-free

Tests + CI:

• tests/test_semantic_coverage.py: principles have guards, constraints have direct guard mapping, habitability guards attached to runtime flow, real examples are warning-free
• CI: strict validation for metadata.fictional_example==false, plus pytest

Result: community-space-brand now passes with 0 errors / 0 warnings.