feat(p1x): de-hardcode VLM models/endpoints to vlm_config (DGX-ready)

Migre les call-sites VLM serveur vers la configuration centrale pour
fonctionner sur DGX (tunnel Ollama 11434), où gemma4:* est absent et le
port Docker 11435 est mort.

- task_planner, replay_verifier, domain_context, ir_builder, resolve_engine
  (popup): modele -> vlm_config.get_vlm_model(), defaut 11435 -> 11434
  (override GEMMA4_PORT legacy conserve)
- resolve_engine (grounding bbox x2): nouvel helper
  vlm_config.get_bbox_grounding_model() (var dediee RPA_BBOX_GROUNDING_MODEL,
  fallback RPA_GROUNDING_MODEL puis qwen2.5vl:7b-rpa) -> desambiguise le
  conflit D5-v3b, bbox_2d + num_ctx 4096 preserves
- safety_checks_provider: defaut -> get_vlm_model(), override
  RPA_SAFETY_CHECKS_LLM_MODEL preserve
- ui_detector: default_factory + resolution lazy (corrige aussi un gel a
  l'import), pas d'appel reseau a l'import
- field_extractor: property lazy via vlm_config

TDD strict (RED->GREEN), 305 tests verts, tests mockes HTTP (zero dependance
DGX reel), aucun alias Ollama.

Hors perimetre (arbitrage Dom): client Lea agent_v1/executor.py (gele),
chemin V4 observe_reason_act (RPA_REASONING_MODEL), core/config.py defaults.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

tests/unit/test_extraction_engine.py

@@ -311,6 +311,26 @@ class TestDataStore:
 
 class TestFieldExtractor:
 
+    def test_model_default_via_vlm_config(self, monkeypatch):
+        """Sans modèle explicite, ollama_model est résolu via vlm_config (lazy, pas gemma4 en dur)."""
+        monkeypatch.delenv("RPA_VLM_MODEL", raising=False)
+        monkeypatch.delenv("VLM_MODEL", raising=False)
+        from unittest.mock import patch
+        import core.extraction.field_extractor as fe_mod
+
+        extractor = FieldExtractor()
+        with patch.object(fe_mod.vlm_config, "get_vlm_model", return_value="modele-resolu:test"):
+            assert extractor.ollama_model == "modele-resolu:test"
+
+    def test_explicit_model_preserved(self):
+        """Un modèle explicite est conservé (résolution non déclenchée)."""
+        from unittest.mock import patch
+        import core.extraction.field_extractor as fe_mod
+        extractor = FieldExtractor(ollama_model="mon-modele:pin")
+        with patch.object(fe_mod.vlm_config, "get_vlm_model",
+                          side_effect=AssertionError("ne doit pas être appelé")):
+            assert extractor.ollama_model == "mon-modele:pin"
+
     def test_extract_file_not_found(self, sample_schema):
         extractor = FieldExtractor()
         result = extractor.extract_fields("/tmp/nonexistent.png", sample_schema)