[WIP] RFC: Scaling the knowledge graph data service

[baiirun]

Will eventually be too computation and storage intensive for a single indexer to index all of the spaces.

Working ideas

• Shard indexing by space. This includes all subspaces of this "root node" and all of their subspaces as well.
• Subspace service to track all subspaces associate with a space. This effectively creates a list from a DAG of space->subspaces and returns the list in an API. This is required by the indexer to know which subspaces it should index as a side-effect of which space it chooses to index.
• Block lists are curated by each space. The indexer needs to know the blocklist before it starts indexing. A block list service can aggregate the list at a space and space-tree level. Indexer queries this before starting indexing.
• Knowledge Graph Entity Service. Stores all entities, their ids, their names, and their spaces