refactor: consolidate LLM JSON parsing into shared llm_utils module

Extracted duplicate _strip_markdown_fences / _parse_llm_json functions
from 7 files into app/services/llm_utils.py. Two shared functions:
- strip_markdown_fences(): fence stripping only
- parse_llm_json(): fence stripping + JSON parse + error logging

Files updated: flowpilot_engine, knowledge_flywheel, session_to_flow_service,
ai_tree_generator_service, ai_fix_service, ai_chat_service, kb_conversion_service

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

backend/app/services/session_to_flow_service.py

@@ -5,7 +5,6 @@ flow with fallback branches, powered by AI.
 """
 import json
 import logging
-import re
 import uuid
 from typing import Any, Optional
 from uuid import UUID
@@ -16,6 +15,7 @@ from sqlalchemy.ext.asyncio import AsyncSession
 from app.core.ai_provider import get_ai_provider
 from app.core.config import settings
 from app.core.ai_tree_validator import validate_generated_procedural_steps
+from app.services.llm_utils import parse_llm_json
 from app.models.session import Session
 from app.models.tree import Tree
 
@@ -80,13 +80,6 @@ Rules:
 """
 
 
-def _strip_markdown_fences(text: str) -> str:
-    """Strip markdown code fences if the model wrapped its JSON response."""
-    text = text.strip()
-    match = re.match(r"^```(?:json)?\s*([\s\S]*?)```$", text)
-    if match:
-        return match.group(1).strip()
-    return text
 
 
 def _build_session_context(session: Session, tree: Optional[Tree]) -> str:
@@ -222,11 +215,7 @@ async def generate_flow_from_session(
     )
 
     # Strip markdown fences and parse JSON
-    raw_text = _strip_markdown_fences(raw_text)
-    try:
-        generated = json.loads(raw_text)
-    except json.JSONDecodeError as e:
-        raise ValueError(f"AI returned invalid JSON: {e}") from e
+    generated = parse_llm_json(raw_text)
 
     # Validate the generated steps
     val_errors = validate_generated_procedural_steps(generated)