fix: resolve 78 test failures via shared stub cleanup and dependent fixes (#85) (#88)
Centralize and enhance test stubs to implement the full CommanderGameState
interface (battlefield(), living_players(), stack_is_empty(), sim_state,
active_player(), priority_player(), to_dict()). Fix all dependent test
failures across 12 files:
- conftest.py: Enhanced StubCard/StubPlayer/StubGameState with full API
compatibility; added basic_creature fixture; fixed fresh_app singleton leak
- test_phase3.py: Added battlefield(), active_player(), prompt_builder compat
- test_integration.py: Fixed ThreatAssessor import (class→function), added
threat_fn dict wrapper, sim_state delegate
- test_phase6.py/test_phase8.py: Replaced deprecated get_event_loop() with
asyncio.run()
- test_benchmark.py: Same async fix + ThreatAssessor functional API
- sim/test_engine.py: Removed nonexistent seed= param, use run() method
- sim/test_models.py: basic_creature fixture now in conftest
- deck_builder/test_models.py: Updated ValidationError regex match
- state_logger.py: Added monotonic counter for unique snapshot filenames
- ai_opponent.py: Moved _CANNED dict to module-level constant (dataclass fix)
- web/app.py: Mount lobby router (fixes 6 LobbyAPI test failures)
- test_phase5.py: SessionStore + api._store reset fixture
Result: 76 failures + 2 errors → 0 failures, 0 errors (331 passed, 30 skipped)
Co-authored-by: Claude <claude@anthropic.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
