fix: guard against empty/filtered LLM responses in OpenAI extension

[qizwiz]

What

Adds a null check in extensions/llms/openai/pandasai_openai/base.py before returning response.choices[0].message.content.

Why

The OpenAI SDK can return two silent failure modes that cause unhandled crashes at this line:

1. IndexError — choices is an empty list ([]) when the provider returns no completions
2. AttributeError — choices[0].message is None when content is filtered (e.g. Gemini returns HTTP 200 with PROHIBITED_CONTENT finish reason)

Fix

if not response.choices or response.choices[0].message is None:
    raise ValueError("LLM returned empty or filtered response")
return response.choices[0].message.content

Detected by pact static analysis.