feat(isolation): DynamicVRAM compatibility for process isolation

DynamicVRAM's on-demand model loading/offloading conflicted with  process isolation in three ways: RPC tensor transport stalls from mid-call GPU offload, race conditions between model lifecycle and active RPC operations, and false positive memory leak detection from changed finalizer patterns.

- Marshal CUDA tensors to CPU before RPC transport for dynamic models
- Add operation state tracking + quiescence waits at workflow boundaries
- Distinguish proxy reference release from actual leaks in cleanup_models_gc
- Fix init order: DynamicVRAM must initialize before isolation proxies
- Add RPC timeouts to prevent indefinite hangs on model unavailability
- Prevent proxy-of-proxy chains from DynamicVRAM model reload cycles
- Add torch.device/torch.dtype serializers for new DynamicVRAM RPC paths
- Guard isolation overhead so non-isolated workflows are unaffected
- Migrate env var to PYISOLATE_CHILD

pyproject.toml

@@ -10,6 +10,25 @@ homepage = "https://www.comfy.org/"
 repository = "https://github.com/comfyanonymous/ComfyUI"
 documentation = "https://docs.comfy.org/"
 
+[tool.comfy.host]
+allow_network = false
+writable_paths = ["/dev/shm", "/tmp"]
+
+[tool.comfy.host.whitelist]
+"ComfyUI-Crystools" = "*"
+"ComfyUI-Florence2" = "*"
+"ComfyUI-GGUF" = "*"
+"ComfyUI-KJNodes" = "*"
+"ComfyUI-LTXVideo" = "*"
+"ComfyUI-Manager" = "*"
+"comfyui-depthanythingv2" = "*"
+"comfyui-kjnodes" = "*"
+"comfyui-videohelpersuite" = "*"
+"comfyui_controlnet_aux" = "*"
+"rgthree-comfy" = "*"
+"was-ns" = "*"
+"websocket_image_save.py" = "*"
+
 [tool.ruff]
 lint.select = [
   "N805", # invalid-first-argument-name-for-method