docs(step-8): aggiungi regole per parametri ottimali

fix(step-9): passa SYSTEM_PROMPT come campo system nell'API Ollama
anziche concatenato nel prompt — risolve risposte di fallback errate
con modelli piccoli

step-9/rag.py

@@ -60,10 +60,11 @@ def embed(text: str) -> list[float]:
 
 # ─── Generazione ──────────────────────────────────────────────────────────────
 
-def call_ollama(prompt: str) -> str:
+def call_ollama(prompt: str, system: str = "") -> str:
     """Chiama Ollama /api/generate e ritorna la risposta."""
     payload = json.dumps({
         "model": LLM_MODEL,
+        "system": system,
         "prompt": prompt,
         "stream": False,
         "think": not NO_THINK,
@@ -110,6 +111,7 @@ def retrieve(collection: chromadb.Collection, question: str) -> list[dict]:
 # ─── Prompt ───────────────────────────────────────────────────────────────────
 
 def build_prompt(question: str, chunks: list[dict]) -> str:
+    """Ritorna (system, user_prompt) separati per l'API Ollama."""
     context_parts = []
     for i, c in enumerate(chunks, start=1):
         header = f"[Contesto {i}"
@@ -121,11 +123,8 @@ def build_prompt(question: str, chunks: list[dict]) -> str:
         context_parts.append(f"{header}\n{c['text']}")
 
     context = "\n\n".join(context_parts)
-    return (
-        f"{SYSTEM_PROMPT}\n\n"
-        f"{context}\n\n"
-        f"Domanda: {question}"
-    )
+    user_prompt = f"{context}\n\nDomanda: {question}"
+    return SYSTEM_PROMPT, user_prompt
 
 
 # ─── Loop interattivo ─────────────────────────────────────────────────────────
@@ -148,10 +147,10 @@ def answer(question: str, collection: chromadb.Collection, verbose: bool) -> Non
             print(f"      {c['text'][:120].replace(chr(10), ' ')}...")
         print("──────────────────────────────────────────────────────────────\n")
 
-    prompt = build_prompt(question, chunks)
+    system, prompt = build_prompt(question, chunks)
 
     try:
-        response = call_ollama(prompt)
+        response = call_ollama(prompt, system=system)
     except (urllib.error.URLError, OSError) as e:
         print(f"❌ Errore generazione: {e}")
         return