Kernel crash from .fit_transform

Nobody has claimed this yet.

I am trying to run the BERTopic model on a Mac (16 GB RAM, Intel i7) using example data from sklearn.datasets (fetch_20newsgroups). The following lines of code are run without any problems.

```
topic_model = BERTopic()
docs = fetch_20newsgroups(subset='all', remove=('headers', 'footers', 'quotes'))['data']
```

But when running this line the kernel crashes:
```
topics, probs = topic_model.fit_transform(docs)
```
The problem seems to occur towards the end, because it happens after 100 % of the documents are transformed into embeddings and it says embedding - completed. There is the following warning message before the kernel dies:
```
OMP: Info #276: omp_set_nested routine deprecated, please use omp_set_max_active_levels instead.
UserWarning: resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown
warnings.warn('resource_tracker: There appear to be %d '

Restarting kernel...
```

The same happens when running it on just a few and short documents (even as short as sentences).

Do you know what the problem might be and how to fix it?

Start by reproducing the crash with BERTopic(), fetch_20newsgroups, and topic_model.fit_transform(docs), including the short-document case. Examine the OMP and resource_tracker warnings and the point after embedding completes; done means fit_transform finishes without restarting the kernel.

Written by the indexing model from the issue text.