feat(M5-M7): embed canonical material names in USD via customData + pxr direct read

- export_step_to_usd.py: accept --material_map CLI arg, write
  schaeffler:canonicalMaterialName as customData on each Mesh prim,
  fix geometry transform (strip shape Location before face exploration,
  apply both face_loc and shape_loc sequentially)
- import_usd.py: after Blender USD import, use pxr to read customData
  directly from the USD file — builds {part_key: material_name} lookup
  (Blender ignores STRING primvars and customData, but pxr reads both)
- _blender_materials.py: add apply_material_library_direct() for exact
  dict-based material assignment without name-matching heuristics
- _blender_scene_setup.py: prefer direct USD lookup, fall back to
  name-matching for legacy USD files without material metadata
- export_glb.py (generate_usd_master_task): resolve material_map via
  material_service.resolve_material_map() and pass to subprocess;
  include material hash in cache key for invalidation
- ROADMAP.md: update P5 status, add M5-M7 milestones

Tested: 3/3 parts matched (ans_lfs120), 172/175 parts matched
(F-802007.TR4-D1-H122AG). Previous: 0/25 matched.

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

.claude/commands/implement.md

@@ -62,7 +62,7 @@ docker compose exec frontend npx tsc --noEmit
 
 ### Celery / Tasks
 - `step_processing` queue: fast tasks only (< 5s) — metadata extraction, dispatch
-- `thumbnail_rendering` queue: ALL Blender/render-worker calls — **never queue Blender on step_processing**
+- `asset_pipeline` queue: ALL Blender/render-worker calls — **never queue Blender on step_processing**
 - Task location: `backend/app/domains/pipeline/tasks/` — not `backend/app/tasks/`
 - `step_tasks.py` is a 23-line shim — do not add logic there
 - Write `self.request.id` to `render_job_doc.celery_task_id` at task start (for cancellation)
@@ -91,7 +91,7 @@ No HTTP blender-renderer service — everything goes through Celery:
 step_processing queue:
   backend/app/domains/pipeline/tasks/extract_metadata.py  (OCC parsing)
 
-thumbnail_rendering queue (render-worker container):
+asset_pipeline queue (render-worker container):
   backend/app/domains/pipeline/tasks/render_thumbnail.py
   → subprocess: render-worker/scripts/export_step_to_gltf.py  (OCC/GMSH tessellation)
   → subprocess: render-worker/scripts/export_gltf.py  (Blender: materials, seams, sharp)