Adding a pipeline component for zero shot LLM document classification and NER

[JaMe76]

Enhancement 🚀
With support of LLM integration libraries like Llama index or Langchain it is fairly easy to add a pipeline component for zero shot classification or entity recognition using LLMs.

Motivation 💪
Everybody knows what LLMs are capable of. The approach has the advantage that no model has to be trained.

The enhancement is not intended to allow more complex integrations with LLMs... There are much better library for that.
However, it could be useful to add this zero shot classification component, e.g. if somebody wants to load layout chunks into a database/vector store with additional meta data classes.

Alternatives ⚖️
NN

Additional context 🧬
Screenshots, etc. if relevant