feat(@projects/@magic-civilization): ✨ update combat_balance extraction status

Co-Authored-By: Lilith Autocommit <noreply@atlilith.com>

.project/objectives/p2-65-extract-mc-state-crate.md

@@ -182,10 +182,39 @@ Estimated: ~2 hours.
 
 Pre-existing 1-day estimate did not account for either finding. Revised:
 
-- Phase 0a — Resolve `CombatBalance` duplication (~2 hr).
+- Phase 0a — Resolve `CombatBalance` duplication (~2 hr). **✓ Done 2026-05-09.**
 - Phase 0b — Move `ScoringWeights` group to mc-core (~1-2 hr).
 - Phase 1-7 — Original mc-state extraction (~1 day).
-- **Total: 1.5-2 days of focused work.**
+- **Remaining: ~1 day of focused work.**
+
+## Phase 0a — done 2026-05-09
+
+Unified `CombatBalance` into `mc-core::combat_balance::CombatBalance`
+(8-field struct: 5 JSON-canonical + 3 mc-ai-internal scoring fields).
+
+- New file: `src/simulator/crates/mc-core/src/combat_balance.rs`. Public
+  re-export at `mc_core::CombatBalance` and `mc_core::parse_combat_balance`.
+- `src/simulator/crates/mc-turn/src/combat_balance.rs` now a thin shim:
+  re-exports `mc_core::combat_balance::CombatBalance`, retains
+  `load_combat_balance` wrapper that maps `serde_json::Error` →
+  `crate::end_conditions::ConfigError`.
+- `src/simulator/crates/mc-ai/src/policy.rs` — local `CombatBalance` +
+  `Default` impl deleted; `pub use mc_core::CombatBalance;` added at
+  module head. `mc_ai::CombatBalance` re-export from `lib.rs` continues
+  to work transparently (re-export of re-export).
+- **Default-value bug fix**: previous `mc-turn` defaults disagreed with
+  the canonical `combat_balance.json` on three fields. Defaults now
+  match JSON exactly: `denial_value_factor 0.5 → 0.6`,
+  `capture_civilian_xp_award 25 → 0`,
+  `destroy_civilian_xp_award_multiplier 0.5 → 1.0`. Production behaviour
+  was already correct (JSON wins on load); this aligns the
+  deserialise-from-empty / `::default()` path with canonical content.
+- Verification on apricot:
+  - `cargo check -p mc-core -p mc-ai -p mc-turn` ✓ clean.
+  - `cargo test -p mc-core combat_balance` — 4/4 ✓.
+  - `cargo test -p mc-ai --test capture_scoring` — 8/8 ✓.
+  - `cargo test -p mc-turn --test ransom` — 12/12 ✓.
+  - `cargo test -p mc-turn --lib combat_balance` — 2/2 ✓.
 
 Given the cross-cutting nature (touches mc-core, mc-ai, mc-turn,
 mc-balance, all consumer crates, both API surfaces, plus 5+ test files)