add sources field to MetaEdge so that infores identifiers are returne…

[sierra-moxon]

This PR adds the sources field to MetaEdge so that infores identifiers are returned in the meta knowledge graph.

Recently, the modeling and TAQA teams were presented with the use case:
"Find me all the unique triples provided to Translator by SEMMEDDB"

We know that SEMMEDDB is currently being served by both BTE and RTX-KG2. Some questions we wanted to ask with code:

• Are there any other SEMMEDDB edges in Translator not from these two aggregators?
• What are the unique SPOQs that we can expect BTE to provide from SEMMEDDB, how about RTX-KG2?
• Do SPOQs from RTX-KG2 (and ultimately SEMMEDDB) match those provided from BTE (via SEMMEDDB)?

Note: we also have several instances of Chembl and clinicaltrials.gov being ingested by multiple teams. It would be handy for the modeling team to be able to compare translations of these important sources between groups (with the ultimate goal of having one transform schema per ingest that groups could follow to keep ingests consistent).

If testing software finds edges in the metakg that need to change with Biolink releases, it would be very handy for QC to be able to point to the primary_knowledge_source (or sources) for more information.

If not for 1.4, then perhaps this could be considered for the next release.

[edeutsch]

I think this is a reasonable addition, but I question if we can tolerate to make a mandatory (and thus breaking) addition at this late stage of 1.4. I would be happier with making it optional in 1.4 and mandatory in 1.5 is we so desire.