Add full multi-step decision workflow for SINGLE_UNCAPPED and DOUBLE_UNCAPPED
outcomes, replacing the previous stub that fell through to basic single/double
advancement. The decision tree follows the same interactive pattern as X-Check
resolution with 5 phases: lead runner advance, defensive throw, trail runner
advance, throw target selection, and safe/out speed check.
- game_models.py: PendingUncappedHit model, 5 new decision phases
- game_engine.py: initiate_uncapped_hit(), 5 submit methods, 3 result builders
- handlers.py: 5 new WebSocket event handlers
- ai_opponent.py: 5 AI decision stubs (conservative defaults)
- play_resolver.py: Updated TODO comments for fallback paths
- 80 new backend tests (2481 total): workflow (49), handlers (23), truth tables (8)
- Fix GameplayPanel.spec.ts: add missing Pinia setup, fix component references
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Fix critical encoding mismatch where calculate_on_base_code() returned
bit-field encoding (3=R1+R2, 4=R3) but runner_advancement.py charts
expected sequential encoding (3=R3, 4=R1+R2). Values 3 and 4 were
swapped, causing wrong groundball results for R1+R2 and R3-only scenarios.
Add comprehensive test coverage:
- 1184 invariant tests (structural correctness across all outcomes × base codes)
- 49 hit truth table tests (SINGLE_1/2, DOUBLE_2/3, TRIPLE, HOMERUN)
- 33 walk truth table tests (WALK, HBP with stat flag verification)
- 42 simple out truth table tests (STRIKEOUT, LINEOUT, POPOUT, WP, PB)
- 88 groundball truth table tests (GB_A/B/C × infield back/in/corners_in × locations)
Total: 2401 unit tests passing.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
