chore: unpin llama-cpp-python

[anakin87]

Related Issues

• Starting from the release of llama-cpp-python==0.3.0, integrations tests were running for > 4 hours.
• In fix: pin llama-cpp-python<0.3.0 #1111, I pinned the package to a previous version.
• Since the issues were not solved with the release of llama-cpp-python==0.3.1, I spent some time debugging the problem.

Proposed Changes:

• make the hanging test run in a reasonable amount of time
  • force the choice of the tool (as shown in the docs)
  • also set n_ctx to 512 (default in llama-cpp-python, while the default in our integration is 0 = inferred from the model)
• unpin llama-cpp-python

How did you test it?

CI

Notes for the reviewer

I think in the future we will rework function calling with llama-cpp-python and remove the test with this model (functionary-small-v2.4), which is becoming obsolete.
In the meantime, I prefer to try to be compatible with the latest versions of this library.

Checklist

• I have read the contributors guidelines and the code of conduct
• I have updated the related issue with new insights and changes
• I added unit tests and updated the docstrings
• I've used one of the conventional commit types for my PR title: fix:, feat:, build:, chore:, ci:, docs:, style:, refactor:, perf:, test:.

[julian-risch]

LGTM! 👍