[Bug] MiniMaxH3: DynamicVRAM causes infinite hang at "Model Initializing..." on RTX 4070 12GB; custom nodes surface underlying CUDA illegal memory access

### Custom Node Testing

- [x] I have tried disabling custom nodes and the issue persists (see [how to disable custom nodes](https://docs.comfy.org/troubleshooting/custom-node-issues#step-1%3A-test-with-all-custom-nodes-disabled) if you need help)

### Expected Behavior
MiniMaxH3 should begin sampling and generate video when loaded with DynamicVRAM enabled on a 12GB NVIDIA GPU.

### Actual Behavior
The process hangs indefinitely at `0%| 0/20 [00:00

- **Two failure modes, same root cause:** Without custom nodes = silent hang. With custom nodes = `cudaErrorIllegalAddress` crash. The custom nodes are not the cause; they perform `soft_empty_cache` / `mem_get_info` calls that surface an already-corrupt CUDA context.
- This appears to be the same underlying issue as #14573 (hang at Model Initializing with DynamicVRAM on low-VRAM NVIDIA), #15488 (GPU lost / illegal memory access with MiniMax H3 on high-RAM Windows systems), and #15352 (AIMDO crash post-unload with MiniMax H3).
- The fact that `CUDA_LAUNCH_BLOCKING=1` does **not** change the behavior rules out a simple async kernel race condition. The corruption appears to happen during the initial VBAR/stream setup or weight prefetch phase before any kernel even executes.