SPIKE: Investigate harvest and index bottleneck

[ghost]

No description provided.

[ghost]

Here are my suggestions thus far.

1. scale pods configuring each instance to index a specific class model, or at least divide up class models between a few instances
2. explore mechanism to combine granular sparql constructs or process to combine them without breaking changes
3. implement debug logging identifying duplicate triples in sparql responses

[ghost]

There is a potentially significant optimization that can be made for all top-level properties of the scholar's domain model. A single SparQL construct query can be submitted for all individuals for a property at once. These can be cached in a map for lookup while filling up properties of each individual's Solr document.