Check, for documents uploaded originally in BDOC/GATE JSON format, are exported correctly as BDOC/GATE JSON format

[twinkarma]

The option to export as GATE format converts the id field to name and any field in the root that is not named text are placed into a feature field. This can cause problems if the document was originally uploaded as GATE format:

Original document:

{
  "name": 32,
  "text": "Document text",
  "features": {
    "text2": "Document text 2",
    "feature1": "Feature text"
  },
  "offset_type":"p",
  "annotation_sets": {...}
}

Incorrect output, notice the features, offset_type and annotation_sets fields are placed inside the root feature field:

{
  "name": 32,
  "text": "Document text",
  "features": {
    "features": {
      "text2": "Document text 2",
      "feature1": "Feature text"
    },
   "offset_type":"p",
   "annotation_sets": {...}
  },
  "annotation_sets": {...}
}

Where the correct output should be:

{
  "name": 32,
  "text": "Document text",
  "features": {
    "text2": "Document text 2",
    "feature1": "Feature text"
  },
  "offset_type":"p",
  "annotation_sets": {...}
}

Clarifications on how the annotation_sets field should be merged are in #348