Empty Sana Latent Image AttributeError: 'EmptySanaLatentImage' object has no attribute 'device'

### 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

## Expected Behavior

The `Empty Sana Latent Image` node should correctly create a latent tensor on the active device (CUDA/XPU/CPU) without throwing errors.

Expected behavior:

* The node should initialize the latent tensor successfully.
* It should work with Intel XPU devices and other supported backends.
* Workflow execution should continue normally.

## Current Issue

Running the node causes:

```python
AttributeError: 'EmptySanaLatentImage' object has no attribute 'device'
```

The issue occurs in:

```python
latent = torch.zeros([batch_size, 32, height // 32, width // 32], device=self.device)
```

because `self.device` is never initialized inside `EmptySanaLatentImage`.

## Suggested Fix

Replace:

```python
device=self.device
```

with proper device detection such as:

```python
device = torch.device("xpu" if torch.xpu.is_available() else "cuda" if torch.cuda.is_available() else "cpu")
```

or use ComfyUI device management utilities.

### Actual Behavior

When executing the workflow with the Empty Sana Latent Image node, ComfyUI crashes with the following error:

AttributeError: 'EmptySanaLatentImage' object has no attribute 'device'

The error occurs during latent tensor creation inside nodes.py:

latent = torch.zeros([batch_size, 32, height // 32, width // 32], device=self.device)

Since self.device is never defined or initialized in EmptySanaLatentImage, the node fails immediately during execution.

Environment
ComfyUI Version: 0.22.1
Python: 3.13.12
PyTorch: 2.12.0+xpu
GPU: Intel Arc 130V
Backend: XPU
OS: Windows

### Steps to Reproduce

>

### Debug Logs

```powershell
>
```

### Other

_No response_