fix: upgrade deprecated openai api and model davinci to gpt-3.5-turbo

[lucumango]

Updated the Synthesize.ipynb research notebook to use the modern openai>=1.0.0 Python client and migrating the text generation away from deprecated models.

The davinci model and the openai==0.28 library are currently deprecated and throw an InvalidRequestError/deprecation warning when researchers attempt to run the data synthesis notebook.

Summary:

• Updated !pip install openai==0.28 to !pip install "openai>=1.0.0" pandas.
• Removed the hardcoded assignment of openai.api_key. The script now securely initializes the client using the OPENAI_API_KEY from the system's environment variables (os.environ.get).
• Replaced the deprecated openai.Completion.create utilizing the davinci engine with the modern client.chat.completions.create syntax using gpt-3.5-turbo.
• Wrapped the API call in a basic try/except block to gracefully handle potential API rate limits or quota errors without crashing the entire batch loop.
• Verified syntax against the new openai v1.x client specifications.

Fixes #21

[lucumango]

More on deprecations: https://developers.openai.com/api/docs/deprecations/