Standardize Finder, Readers, and Retriever interfaces

[tanaysoni]

This PR makes the following changes:

Retriever-Reader Interface

• in the current implementation, Retrievers return a list of texts and their associated meta_data. This PR introduces a Document schema that encapsulates documents and their meta_data. It simplifies the Retriever-Reader pipeline since now Retrievers only need to return a list of Document.

Simplify IDs

• the document_id field(Document.id) is now the _id field from ES
• external_source_id is added to identify the source file for the case when a large file is split across multiple ES documents.

Code Style

• introduce Pydantic and Python type hints

[tholor]

Looking great. Just minor comment about doc string.