Thinking models in ollama exit early

Nobody has claimed this yet.

### Have you searched existing issues? 🔎

- [x] I have searched and found no existing issues

### Desribe the bug

Even when passing in 'stop': None to bertopic.OpenAI, "stop": "\n" gets passed into the openai model call, which causes thinking models to exit early in Ollama (in particular, gemma4:e2b)

### Reproduction

```python
from bertopic import BERTopic
import openai
from bertopic.representation import OpenAI

client = openai.OpenAI(
base_url="http://localhost:11435/v1",
api_key="ollama"
)

representation_model = OpenAI(client, model="gemma4:e2b", generator_kwargs={"stop": None})
```

### BERTopic Version

0.17.4

Start by reproducing the example with BERTopic 0.17.4, Ollama, and gemma4:e2b, then trace how bertopic.representation.OpenAI handles generator_kwargs and forwards the stop option to the OpenAI-compatible call. Done means passing stop=None no longer sends a newline stop value and thinking models do not exit early; no file or test is named in the issue.

Written by the indexing model from the issue text.