in ARAX, in the KG, handle case of attribute_source being null

[saramsey]

This issue is motivated by #2702

Sometimes a KP returns an edge to ARAX expand that has an edge attribute with attribute_source: null, like this:

            {
              "attribute_source": null,
              "attribute_type_id": "biolink:Attribute",
              "attributes": null,
              "description": null,
              "original_attribute_name": "hetio_source",
              "value": [
                "DrugBank (enzyme)"
              ],
              "value_type_id": null,
              "value_url": null
            }

In that case, ARAX should just put, in place of null, in the edge attribute's attribute_source slot, the infores for the KP (e.g., infores:automat-hetionet or whatever).

Furthermore, ARAX-Overlay's NGD module (compute_ngd.py) adds a bunch of edge attributes, and these should have their attribute_source set to infores:arax (per discussion on Slack).

[saramsey]

Need to come up with a simple example query graph that shows this issue, and put the query graph in this issue, before handing off to a team member for implementation.

[saramsey]

Working on this issue in branch issue-2707

[saramsey]

This seems to be a minimum query graph:

{
   "edges": {
      "e00": {
         "subject":   "n00",
         "object":    "n01",
         "predicates": ["biolink:physically_interacts_with"]
      }
   },
   "nodes": {
      "n00": {
         "ids":        ["CHEBI:46195"]
      },
      "n01": {
         "ids":  ["NCBIGene:1571"]
      }
   }
}

produces a single result ("CYP2E1") which has a single edge from infores:automat-hetionet, with this edge attribute in it:

Edge Attribute

{
  "attribute_source": null,
  "attribute_type_id": "biolink:knowledge_level",
  "attributes": null,
  "description": null,
  "original_attribute_name": "knowledge_level",
  "value": "not_provided",
  "value_type_id": null,
  "value_url": null
}

[saramsey]

PR submitted #2708

[bazarkua]

PR merged into master

[saramsey]

All 154 standard pytests passed.