How to map the instance information in the generated BOP dataset to the pose information one by one

### Describe the issue

Dear developers and fellow code users,

I have a question regarding the generation of the BOP instance segmentation dataset using BlenderProc. I noticed that the segmented instances in **scene_gt_coco.json** do not correspond sequentially to the pose instances in **scene_gt.json**. How can I match the instances in **scene_gt_coco.json** to their corresponding pose entries in **scene_gt.json**?

For example, after visual inspection, the second instance in **scene_gt_coco.json** corresponds to the eighth instance in **scene_gt.json**.


### Minimal code example

```python
None
```

### Files required to run the code

_No response_

### Expected behavior

map the instance information in the generated BOP dataset to the pose information one by one

### BlenderProc version

2.8.0

Start by comparing the instance entries in scene_gt_coco.json with the pose entries in scene_gt.json for one generated scene, using the reported second-versus-eighth example as a check. Document the reliable correspondence rule and identify how users can map every segmented instance to its pose entry without relying on visual inspection.

Written by the indexing model from the issue text.