Add the playable scene loop: boot through mission to reward

Wires every autoload up to a real UI: boot -> profile select/create
-> world map (three mission markers, positioned from data, not
hardcoded pixels) -> mission intro (narrative from data/missions/,
a new data type with its own schema) -> rule panel -> exercise host
-> one of the three exercise type scenes -> feedback panel -> reward
screen -> back to the map.

exercise_host is the only piece with real orchestration logic: picks
exercises via ExerciseEngine (now fully implemented, validating at
load like make validate-data does), validates answers through the
pure AnswerValidator/DictationAnalyzer, escalates aid over 3 attempts
via AttemptFlow, feeds DifficultyManager so adaptive difficulty is
actually wired in rather than dead code, and records every attempt
through ProgressManager. Every exercise-type scene (multiple_choice,
sentence_correction, dictation) only emits its answer by signal, per
the architecture invariant.

All child-facing text now comes from data/localization/fr.json via
UIText, not hardcoded in scripts or scenes. Getting that right
required extending SchemaValidator to support "additionalProperties"
as a schema (dynamic keys), which also closed a real gap: save.skills
was previously unvalidated against SkillState's actual shape.

16 new headless tests, including two integration tests that instance
exercise_host.tscn directly and drive it by emitting child-scene
signals, stopping short of the final scene transition (which would
tear down the test runner's own tree). Everything that needs a
screen -- layout at different sizes, touch, virtual keyboard, actual
audio playback (no .ogg files exist yet) -- is listed in the new
docs/testing/manual-checklist.md instead of claimed as verified.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

tests/exercises/test_exercise_host_flow.gd

@@ -0,0 +1,88 @@
+extends GutTest
+## Test d'integration de exercise_host : regle -> exercice -> validation ->
+## enregistrement de la tentative. S'arrete avant la fin de la mission : aller
+## jusqu'au bout declencherait un vrai changement de scene (GameManager.goto),
+## ce qui casserait le processus de test lui-meme.
+
+var _profile_id: String = ""
+
+
+func before_each() -> void:
+    _profile_id = ProfileManager.create_profile("Test", "CE1")
+    ProfileManager.select_profile(_profile_id)
+    GameManager.pending_mission_id = "mission_01_pre_des_mots"
+
+
+func after_each() -> void:
+    ProfileManager.delete_profile(_profile_id)
+    GameManager.pending_mission_id = ""
+
+
+func _start_host() -> Control:
+    var host: Control = load("res://scenes/exercises/exercise_host.tscn").instantiate()
+    add_child_autofree(host)
+    await wait_process_frames(1)
+    return host
+
+
+func test_rule_panel_is_shown_before_any_exercise() -> void:
+    var host := await _start_host()
+    var overlay: Control = host.get_node("%OverlayContainer")
+    assert_eq(overlay.get_child_count(), 1)
+    assert_eq(host.get_node("%ExerciseContainer").get_child_count(), 0)
+
+
+func test_dismissing_rule_panel_loads_the_first_exercise() -> void:
+    var host := await _start_host()
+    var overlay: Control = host.get_node("%OverlayContainer")
+    overlay.get_child(0).continued.emit()
+    await wait_process_frames(1)
+
+    assert_eq(overlay.get_child_count(), 0)
+    assert_eq(host.get_node("%ExerciseContainer").get_child_count(), 1)
+    assert_eq(host.get_current_exercise()["rule_id"], "regle_accord_adjectif")
+
+
+func test_correct_answer_records_a_successful_attempt() -> void:
+    var host := await _start_host()
+    var overlay: Control = host.get_node("%OverlayContainer")
+    overlay.get_child(0).continued.emit()
+    await wait_process_frames(1)
+
+    var exercise_scene: Control = host.get_node("%ExerciseContainer").get_child(0)
+    var correct_index: int = host.get_current_exercise()["correct_answer"]
+    exercise_scene.answer_submitted.emit(correct_index)
+    await wait_process_frames(1)
+
+    assert_eq(overlay.get_child_count(), 1, "le panneau de retour doit s'afficher")
+    var skill_state := ProgressManager.get_skill_state("accord_groupe_nominal")
+    assert_eq(skill_state["recent_results"], [true])
+    assert_eq(skill_state["recent_hint_used"], [false])
+
+
+func test_wrong_answer_escalates_hint_then_explanation_then_reveal() -> void:
+    var host := await _start_host()
+    var overlay: Control = host.get_node("%OverlayContainer")
+    overlay.get_child(0).continued.emit()
+    await wait_process_frames(1)
+
+    var exercise: Dictionary = host.get_current_exercise()
+    var wrong_index := 0 if exercise["correct_answer"] != 0 else 1
+
+    for _i in 3:
+        var exercise_scene: Control = host.get_node("%ExerciseContainer").get_child(0)
+        exercise_scene.answer_submitted.emit(wrong_index)
+        await wait_process_frames(1)
+        var feedback: Control = overlay.get_child(overlay.get_child_count() - 1)
+        feedback.continue_pressed.emit()
+        await wait_process_frames(1)
+
+    # Apres 3 echecs, l'exercice est termine (reponse montree) et un nouveau
+    # commence : le compteur de tentative revient a 1, et une mission n'est
+    # pas encore terminee (longueur 3, un seul exercice traite ici).
+    assert_eq(host.get_attempt_number(), 1)
+    assert_eq(host.get_completed_count(), 1)
+
+    var skill_state := ProgressManager.get_skill_state("accord_groupe_nominal")
+    assert_eq(skill_state["recent_results"], [false, false, false])
+    assert_eq(skill_state["recent_hint_used"], [false, true, true])