Refactor embedding operations and improve configuration

[patricktrainer]

Fixes #2

Refactor code to use EmbeddingKey class for embedding operations.

• Introduce EmbeddingKey class:

  • Add EmbeddingKey class to encapsulate text and model in src/operations.py.
  • Update functions in src/operations.py to use EmbeddingKey instead of tuples for keys.
  • Refactor write_embedding_to_table, is_key_in_table, list_keys_in_table, and get_embedding_from_table to use EmbeddingKey.

• Refactor embedding operations:

  • Create EmbeddingOperations class in src/embedding_operations.py to encapsulate embedding-related operations.
  • Move functions pickle_embeddings, duckdb_embeddings, and get_similarity into EmbeddingOperations class.
  • Add methods for loading and saving embeddings, checking if an embedding exists, and calculating similarity.

• Update embedding functions:

  • Refactor pickle_embeddings and duckdb_embeddings functions in src/embedding.py to use EmbeddingKey.

• Add error handling:

  • Handle potential exceptions when creating embeddings and provide meaningful error messages in src/openai_client.py.
  • Add error handling for database operations to manage connection issues or query failures in src/connection.py.

• Add configuration file:

  • Add config.yaml with database name, model name, paths to files, OpenAI API key, and number of top documents.

• Add tests:

  • Create unit tests for EmbeddingOperations class in tests/test_embedding_operations.py.
  • Write tests for loading and saving embeddings, checking if an embedding exists, and calculating similarity.
  • Write tests to verify that the application correctly reads and applies settings from the config.yaml file in tests/test_config.py.