refactor(phase3): remove dead services + STL remnant cleanup

Phase 3.2 — Delete orphaned service directories:
  - blender-renderer/ (HTTP microservice replaced by render-worker subprocess)
  - threejs-renderer/ (replaced by render-worker)
  - flamenco/ (removed in migration 032, directory still existed on disk)

Phase 3.2 — Remove STL workflow remnants:
  - analytics.py: remove avg_stl_s from RenderTimeBreakdown schema (always None)
  - kpi_service.py: remove avg_stl_s from return dicts + update docstring
  - frontend/src/api/analytics.ts: remove avg_stl_s from RenderTimeBreakdown interface
  - admin.py: remove dead blender-renderer HTTP configure call (service gone)

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

backend/app/api/routers/admin.py

@@ -366,17 +366,8 @@ async def update_settings(
         await _save_setting(db, k, v)
     await db.commit()
 
-    # Propagate concurrency limit to blender-renderer immediately (no restart needed)
-    if body.blender_max_concurrent_renders is not None:
-        try:
-            import httpx
-            async with httpx.AsyncClient(timeout=3.0) as client:
-                await client.post(
-                    "http://blender-renderer:8100/configure",
-                    params={"max_concurrent": body.blender_max_concurrent_renders},
-                )
-        except Exception:
-            pass  # best-effort; setting is persisted in DB regardless
+    # Note: blender-renderer HTTP service removed; concurrency is now controlled
+    # via render-worker Docker concurrency setting (thumbnail_rendering queue).
 
     return _settings_to_out(await _load_settings(db))
 

backend/app/api/routers/analytics.py

@@ -41,7 +41,6 @@ class ItemStatusBreakdown(BaseModel):
 
 
 class RenderTimeBreakdown(BaseModel):
-    avg_stl_s: Optional[float]
     avg_render_s: Optional[float]
     avg_total_s: Optional[float]
     sample_count: int

backend/app/services/kpi_service.py

@@ -130,8 +130,6 @@ async def render_time_breakdown(
     """Average render duration from completed order lines, scoped to date range.
 
     Uses render_started_at / render_completed_at on order_lines (added in migration 015).
-    avg_stl_s is not tracked at order-line level, so only avg_render_s and sample_count
-    are meaningful here; avg_stl_s is left None for UI compatibility.
     """
     sql = text(
         """
@@ -149,9 +147,8 @@ async def render_time_breakdown(
     result = await db.execute(sql, {"date_from": _parse_date(date_from), "date_to": _parse_date(date_to)})
     row = result.fetchone()
     if row is None or row[1] == 0:
-        return {"avg_stl_s": None, "avg_render_s": None, "avg_total_s": None, "sample_count": 0}
+        return {"avg_render_s": None, "avg_total_s": None, "sample_count": 0}
     return {
-        "avg_stl_s": None,
         "avg_render_s": float(row[0]) if row[0] is not None else None,
         "avg_total_s": float(row[0]) if row[0] is not None else None,
         "sample_count": int(row[1]),