Add model parameter for generating embeddings

[caufieldjh]

This pull request updates the embeddings generation pipeline to support selecting any CurateGPT embedding model, including OpenAI and Hugging Face models, via a new --model parameter. The changes clarify when the OPENAI_API_KEY is required, improve help/documentation, and add comprehensive tests for the new behavior.

Major CLI and API improvements:

• Added a --model (-m) option to the generate-embeddings command, allowing users to specify the CurateGPT embedding model (e.g., openai:text-embedding-3-small or a Hugging Face model). The code now only requires OPENAI_API_KEY if an OpenAI model is selected, and passes the model parameter through all layers to CurateGPT. [1] [2] [3] [4] [5] [6] [7]

Documentation updates:

• Updated the README.md to reflect the new model selection option, clarify requirements for OPENAI_API_KEY, and improve cost explanations. Added usage examples for specifying a model. [1] [2] [3]

Testing improvements:

• Added and updated tests to verify that:
  • The CLI only requires OPENAI_API_KEY for OpenAI models.
  • The model parameter is passed to CurateGPT.
  • The default behavior is preserved if no model is specified.
  • The error handling and help messages are correct for all scenarios. [1] [2] [3] [4]

These changes make the embeddings pipeline more flexible, user-friendly, and robust for both OpenAI and non-OpenAI embedding models.