Improve and complete JSON writing

[pierrepebay]

The goal is to be able to have the JSON outputted by LBAF (when using the internal data structures) to be identical to the input JSON.

Missing elements at this point are (non exhaustive):

• Communications
• subphases
• index
• collection_id
• shared_node data

This issue should also deprecate the use of -1 to read/step through all phases (only explicit enumerations or ranges should be used).

[cwschilly]

@lifflander @ppebay @pierrepebay
I have edited the original comment with more elements that are not being written out. For a couple of these, I am not sure what they represent.

Specifically, what do index and collection_id refer to in the following snippet of an input JSON file?

"id": 0,
      "tasks": [
        {
          "entity": {
            "collection_id": 7,
            "home": 0,
            "id": 3407875,
            "index": [
              12
            ],
            "migratable": true,
            "type": "object"
          },

[ppebay]

@cwschilly I think these quantities are not relevant to LBAF, but I will let @lifflander confirm this (or not)

[ppebay]

N.B.: the first id concerns the phase.

[cwschilly]

I am also not sure what the data in the shared_node block represents:

"metadata": {
  "rank": 0,
  "shared_node": {
    "id": 0,
    "num_nodes": 1,
    "rank": 0,
    "size": 4
  },
  "type": "LBDatafile"
}

For now I am using the approach discussed in the meeting of swallowing everything into a dict and passing it through to the end.

[ppebay]

@pierrepebay regarding the other fields: used in vt-tv, or not?