LTR and RTL annotation for text

Hi,

The annotation tool works great, except the export never state if the start/end are starting from left or right of the text.

Example:

```
[
{
"text": "أجرت وزارة الصحة الإماراتية 123,037 فحصا ضمن خططها لتوسيع نطاق الفحوصات وتكشف عن 1,554 إصابة جديدة بفيروس ",
"id": 3,
"label": [
{
"start": 81,
"end": 86,
"text": "1,554",
"labels": [
"MISC"
]
},
{
"start": 28,
"end": 35,
"text": "123,037",
"labels": [
"MISC"
]
}
],
"annotator": 1,
"annotation_id": 4,
"created_at": "2022-07-12T15:42:19.445022Z",
"updated_at": "2022-07-12T16:05:29.738373Z",
"lead_time": 104.119
},
{
"text": "Donald Trump enters the white house",
"id": 1,
"label": [
{
"start": 0,
"end": 12,
"text": "Donald Trump",
"labels": [
"PER"
]
},
{
"start": 24,
"end": 35,
"text": "white house",
"labels": [
"LOC"
]
}
],
"annotator": 1,
"annotation_id": 3,
"created_at": "2022-07-12T15:32:52.517225Z",
"updated_at": "2022-07-12T15:32:52.517383Z",
"lead_time": 5.122
}
```

The first assume RTL, while the second is LTR (and count accordingly), and the export never state in which direction you're supposed to read the start/end position of entities.