InpaintModelConditioning doesn't work

### Expected Behavior

Using official workflow with inpainting with the flux-fill model, we must have a result according to the prompt.

### Actual Behavior

Using official workflow with inpainting with the flux-fill model, results are not correct.

We have decoding and preview latent and masked region doesn't appear.

### Steps to Reproduce

Load an image in LoadImage.
Edit with mask editor.
Provide a prompt.
Look at the result.

### Debug Logs

```powershell
Run has no problem, but the result is not good.

loaded partially 127.828125 127.583984375 0
torch.Size([1, 1, 1440, 960])
torch.Size([1, 1, 1440, 960])
100%|████████████████████████████████████████████████████| 20/20 [01:09<00:00, 3.46s/it]
Requested to load AutoencodingEngine
0 models unloaded.
loaded partially 64.0 63.999755859375 0
Warning: Ran out of memory when regular VAE decoding, retrying with tiled VAE decoding.
Prompt executed in 73.24 seconds
```

### Other