Sparql query and local setup

[cmdoret]

This PR mainly adds two things (along with a lot of refactoring):

• SPARQL generation: The complete chat workflow is now implemented, including the generation and execution of SPARQL queries.
• local execution: To debug and test more easily, the system can be executed fully locally using an RDF file as knowledge graph, and runnign chroma in client-only mode.

local execution works by setting:

CHROMA_HOST='local'
SPARQL_ENDPOINT='/absolute/path.trig' 
# or './local/path'

It also adds a notebook documenting the whole process. It runs fully locally and can use the OpenAPI API.

Additional improvements:

• Support to embed only a named graph (i.e. a subset of the KG, such as the schema)
• Minimal dataset added in data/test_data.trig

[caviri]

In case it is useful I leave here how I'm running this using a jupyter lab instance.

1. I modify the ENTRYPOINT at the bottom of Dockerfile by ENTRYPOINT ["/bin/bash"]
2. Build the image: docker build kg-llm-interface .
3. Run the image: docker run -it --rm -p 8888:8888 kg-llm-interface
4. Install git: apt get install git
5. Install aikg make install
6. Install Jupyter lab: pip install jupyter lab
7. Add poetry venv as ipykernel: poetry run python -m ipykernel install --user --name aikg
8. Run the JLab session: jupyter lab --allow-root --port 8888 --ip '0.0.0.0'

I have been able to run the ng_sparql.ipynb without errors :)

P.S. Git is required to install aikg when running make install