fix(npc): a move applies at most once per reply

The tag extractor returns every occurrence of a move tag and the validator is
a pure membership test, so MoveApplier applied duplicates. A model repeating a
tag N times therefore picked an amount in unary — the AI choosing a number,
which is the §2 breach the bounded vocabulary exists to prevent.

The currency fix (8e1238c) closed this for denominations only. Generalizing it
to every move that names a thing covers real items too — three give_item(amulet)
tags are one amulet, not three — and lets the currency special case be deleted
rather than kept alongside a second mechanism. ID_MOVES is reused from
MoveValidator instead of restated.

Also caps adjust_disposition on the reply's NET swing. MAX_DELTA was clamped
per tag, so three +15s moved standing by 45 — precisely the wholesale swing its
own comment says it forbids. Applied in emission order, so a later
become_hostile still floors what came before it.

offer_quest and reveal needed no change: add_quest self-guards on has_quest and
mark_revealed is idempotent.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QYa9u7Kdxv5gX4AnwWexy8

client/tests/unit/test_move_applier.gd

@@ -144,18 +144,67 @@ func test_dedup_key_is_move_name_not_just_denomination():
 	assert_eq(gs.purse_copper, 0)
 
 
-func test_repeated_non_currency_give_item_still_applies_each_time():
-	# Pins the scope line: real-item duplicate handling is UNCHANGED by this fix
-	# (it stays bounded only by the cross-reply gifts_given gate, reported separately).
+func test_repeated_give_item_applies_once_for_real_items_too():
 	var gs = GameState.new()
 	MoveApplier.apply([_m("give_item", ["amulet"]), _m("give_item", ["amulet"]), _m("give_item", ["amulet"])],
 		gs, CanonLog.new(), _content(), "fenn")
-	assert_eq(gs.inventory.get("amulet", 0), 3, "non-currency give_item is out of scope for this fix")
+	assert_eq(gs.inventory.get("amulet", 0), 1, "three tags, one amulet")
 
 
-func test_repeated_non_currency_accept_item_still_applies_each_time():
+func test_repeated_accept_item_applies_once_for_real_items_too():
 	var gs = GameState.new()
 	gs.add_item("worn_shortsword", 5)
 	MoveApplier.apply([_m("accept_item", ["worn_shortsword"]), _m("accept_item", ["worn_shortsword"])],
 		gs, CanonLog.new(), _content(), "fenn")
-	assert_eq(gs.inventory.get("worn_shortsword", 0), 3, "non-currency accept_item is out of scope for this fix")
+	assert_eq(gs.inventory.get("worn_shortsword", 0), 4, "two tags, one sword taken")
+
+
+func test_different_real_items_in_one_reply_both_apply():
+	var gs = GameState.new()
+	MoveApplier.apply([_m("give_item", ["amulet"]), _m("give_item", ["ring"])],
+		gs, CanonLog.new(), _content(), "fenn")
+	assert_eq(gs.inventory.get("amulet", 0), 1)
+	assert_eq(gs.inventory.get("ring", 0), 1)
+
+
+func test_repeated_disposition_tags_cannot_exceed_the_per_reply_cap():
+	# MAX_DELTA means "one line cannot swing standing wholesale" (§7 spirit). It was
+	# clamped PER TAG, so three +15 tags moved standing by 45. The cap is per REPLY.
+	var gs = GameState.new()
+	MoveApplier.apply([_m("adjust_disposition", ["+15"]), _m("adjust_disposition", ["+15"]),
+		_m("adjust_disposition", ["+15"])], gs, CanonLog.new(), _content(), "fenn")
+	assert_eq(gs.npc_dispositions.get("fenn", 0), MoveApplier.MAX_DELTA, "+45 is capped to +15")
+
+
+func test_disposition_cap_applies_to_the_net_swing():
+	var gs = GameState.new()
+	MoveApplier.apply([_m("adjust_disposition", ["+10"]), _m("adjust_disposition", ["+10"])],
+		gs, CanonLog.new(), _content(), "fenn")
+	assert_eq(gs.npc_dispositions.get("fenn", 0), 15, "+20 is capped to +15")
+
+
+func test_disposition_tags_of_opposite_sign_net_out():
+	# The cap is on the reply's NET swing, not on the tag count — so a walk-back
+	# inside one reply still lands where the arithmetic says.
+	var gs = GameState.new()
+	MoveApplier.apply([_m("adjust_disposition", ["+10"]), _m("adjust_disposition", ["-4"])],
+		gs, CanonLog.new(), _content(), "fenn")
+	assert_eq(gs.npc_dispositions.get("fenn", 0), 6)
+
+
+func test_become_hostile_still_wins_after_a_disposition_tag():
+	# Moves apply in emission order — the hostile floor must not be undone by
+	# batching the disposition arithmetic.
+	var gs = GameState.new()
+	MoveApplier.apply([_m("adjust_disposition", ["+15"]), _m("become_hostile")],
+		gs, CanonLog.new(), _content(), "fenn")
+	assert_eq(gs.npc_dispositions.get("fenn", 0), MoveApplier.HOSTILE_FLOOR)
+
+
+func test_disposition_cap_is_per_reply_not_per_conversation():
+	# Two separate replies each get a fresh MAX_DELTA budget — standing still moves
+	# over a conversation, it just cannot lurch in a single line.
+	var gs = GameState.new()
+	MoveApplier.apply([_m("adjust_disposition", ["+15"])], gs, CanonLog.new(), _content(), "fenn")
+	MoveApplier.apply([_m("adjust_disposition", ["+15"])], gs, CanonLog.new(), _content(), "fenn")
+	assert_eq(gs.npc_dispositions.get("fenn", 0), 30)