Implement Homeworld mechanics through Caladan reinforcement checkpoint

Connect typed Homeworld custody, transport, battle, income, revival and Caladan reinforcement flows with private multiplayer recovery and controls. Preserve the optimized test workflow and keep unfinished rules modes gated. Verified with 3682 offline tests, typecheck, lint, production build and 40 HTTP tests; existing saved games remain unchanged.

tests/battle-discard-continuations.test.ts

@@ -240,7 +240,7 @@ void test('last real traitor vote stages a mixed-owner public discard after doub
   assert.equal(result.resolvedTreacheryDiscardSequence, 1);
 });
 
-void test('winner hero and loser defense share one mandatory batch before the sole supported casualty allocation', () => {
+void test('loser defense precedes winner casualties and the separate mandatory hero discard', () => {
   const initial = fixture();
   const hero = hold(initial, 'w', 'hero'),
     defense = hold(initial, 'l', 'snooper');
@@ -251,12 +251,17 @@ void test('winner hero and loser defense share one mandatory batch before the so
   assert.equal(c.casualties!.options.length, 1);
   assert.deepEqual(
     pending.pendingTreacheryDiscard!.batch.entries.map((e) => e.discardedBy),
-    ['w', 'l'],
+    ['l'],
   );
+  assert.deepEqual(pending.pendingWinnerDiscards?.cards, [hero]);
+  assert.ok(player(pending, 'w').hand.some((card) => card.id === hero));
   assert.equal(player(pending, 'w').tanks, 0);
   assert.equal(player(pending, 'l').tanks, 5);
   assert.equal(player(pending, 'w').spice, 18);
   const result = resumed(pending);
+  assert.equal(result.resolvedTreacheryDiscardSequence, 2);
+  assert.equal(result.pendingWinnerDiscards, null);
+  assert.ok(!player(result, 'w').hand.some((card) => card.id === hero));
   assert.equal(player(result, 'w').tanks, 2);
   assert.equal(
     result.log.filter((e) => e.automatic?.name === 'Battle casualties').length,
@@ -329,9 +334,13 @@ void test('Moritani reserves loser cards from the mandatory batch and disposes t
   const hero = hold(initial, 'w', 'hero'),
     weapon = hold(initial, 'l', 'projectile'),
     defense = hold(initial, 'l', 'snooper');
-  const pending = frame(
+  const pending = inner(
     beforeLast(initial, { hero, loserWeapon: weapon, loserDefense: defense }),
+    'l',
+    finalAction(),
   );
+  assert.equal(pending.pendingTreacheryDiscard?.continuation.kind, 'winnerMandatoryDiscard');
+  assert.equal(player(pending, 'w').tanks, 2);
   assert.deepEqual(
     pending.pendingTreacheryDiscard!.batch.entries.map((e) => e.card.id),
     [hero],