Not getting annotation.result[N].id while generation prediction using API.

I am using ML backend for detection. Initially, prediction API will return predictions with `annotation.result[N].id.` Now, I am not getting `annotation.result[N].id.` I need this to add results of recognition task.

**Result of old call:**
```
{
"original_width": 1717,
"original_height": 697,
"image_rotation": 0,
"value": {
"x": 7.92712599392253,
"y": 15.064562410329993,
"width": 17.592293634106735,
"height": 3.4433285509325664,
"rotation": 0,
"rectanglelabels": [
"ABC"
]
},
"id": "tofXOU8lTd",
"from_name": "label_rectangles",
"to_name": "image",
"type": "rectanglelabels",
"origin": "prediction",
"score": 0.419921875
}
```

**Result now:**
```
{
"from_name": "label_rectangles",
"score": 0.348876953125,
"to_name": "image",
"type": "rectanglelabels",
"value": {
"height": 3.5868005738880915,
"rectanglelabels": [
"XYZ"
],
"width": 4.48456610366919,
"x": 3.14502038439138,
"y": 80.48780487804879
}
}
```
# Tried And Failed:

- Restart Label studio and ML backend
- Use old working program
- Get prediction of old working projects
- Create annotations from predictions

No repository file, test, or entry point is identified in the issue. Start by reproducing the API prediction response and compare the old and current JSON examples, then trace the prediction-generation path to determine where result IDs are omitted. Done means predictions again include annotation.result[N].id so recognition-task results can be added.

Written by the indexing model from the issue text.