OCR - static pre-annotations not showing when combined with ML backend for inference

**Describe the bug**
A clear and concise description of what the bug is.
static pre-annotations are not showing when used with a ML backend for inference (happens only when "Label All Tasks" is clicked)

**To Reproduce**
Steps to reproduce the behavior:
1. ocr usecase with pre-annotated json tasks (a document image and its task.json : apply pytesseract.image_to_data(IMG) to add boxes and text under prediction)
2. Label config
```

</View>
```
4. connect tesseract ML backend for bbox ocr (only inference), from [here](https://github.com/heartexlabs/label-studio-ml-backend/tree/master/label_studio_ml/examples/tesseract).
5. click "Label All Tasks"
7. static pre-annotations do not appear

**Expected behavior**
A clear and concise description of what you expected to happen.
when I click in "Label All Tasks", static pre-annotations should display, and be able to use it along with the ML Backend (Tesseract here) to create and correct boxes OCR content.

**Screenshots**
If applicable, add screenshots to help explain your problem.

**Environment (please complete the following information):**
- OS: [e.g. iOS]: centos , ubuntu
- Label Studio Version [e.g. 0.8.0]: 1.4.1.post1

**Additional context**
Add any other context about the problem here.
It seems like the ML backend when connected hides static pre-annotations and not able to change which prediction appears first in settings>Machine Learning (happens only when "Label All Tasks" is clicked)
I can give examples in detail if needed for reproduction
Thanks