feat: Neo4j Destination Connector

[ds-filipknefel]

Description

Implements Neo4j destination connector.

Connector takes Unstructured Elements and builds a lexical graph representing relationships between them. It consists of the following entities.

Nodes

• Document - represents the source file elements come from
• UnstructuredElement - represents the Unstructured Element prior to chunking
• Chunk - represents the Unstructured Element post chunking

Edges (Relationships)

• UnstructredElement/Chunk - PART_OF_DOCUMENT -> Document - relationship of belonging to the source file
• UnstructuredElement - PART_OF_CHUNK -> Chunk - relationship between origin elements making up a chunk
• UnstructuredElement - NEXT_ELEMENT -> UnstructuredElement - order of occurrence in the document
• Chunk - NEXT_ELEMENT -> Chunk - order of occurrence in document

Example graph

Notes

• Code used to interact with Neo4j database doesn't utilize apoc add-on library as it is not guaranteed to be available on designated Neo4j instance.
• Ability to build a domain graph - graph representing information from document's contents rather then its structure is not in scope of this PR.

TODO

• Implement upsert support - delete old nodes and relationships when document is re-uploaded
• Add batch splitting
• Add precheck test to integration test suite
• Update descriptions

[rbiseck3]

Don't need this anymore, right?