feat(api): add INCI tool-calling with normalized tool traces

Enable on-demand INCI retrieval in /routines/suggest through Gemini function calling so detailed ingredient data is fetched only when needed. Persist and normalize tool_trace data in AI logs to make function-call behavior directly inspectable via /ai-logs endpoints.

backend/tests/test_ai_logs.py

@@ -0,0 +1,44 @@
+import uuid
+from typing import Any, cast
+
+from innercontext.models.ai_log import AICallLog
+
+
+def test_list_ai_logs_normalizes_tool_trace_string(client, session):
+    log = AICallLog(
+        id=uuid.uuid4(),
+        endpoint="routines/suggest",
+        model="gemini-3-flash-preview",
+        success=True,
+    )
+    log.tool_trace = cast(
+        Any,
+        '{"mode":"function_tools","events":[{"function":"get_product_inci"}]}',
+    )
+    session.add(log)
+    session.commit()
+
+    response = client.get("/ai-logs")
+    assert response.status_code == 200
+    data = response.json()
+    assert len(data) == 1
+    assert data[0]["tool_trace"]["mode"] == "function_tools"
+    assert data[0]["tool_trace"]["events"][0]["function"] == "get_product_inci"
+
+
+def test_get_ai_log_normalizes_tool_trace_string(client, session):
+    log = AICallLog(
+        id=uuid.uuid4(),
+        endpoint="routines/suggest",
+        model="gemini-3-flash-preview",
+        success=True,
+    )
+    log.tool_trace = cast(Any, '{"mode":"function_tools","round":1}')
+    session.add(log)
+    session.commit()
+
+    response = client.get(f"/ai-logs/{log.id}")
+    assert response.status_code == 200
+    payload = response.json()
+    assert payload["tool_trace"]["mode"] == "function_tools"
+    assert payload["tool_trace"]["round"] == 1