Add example for fine tuned Classifier

[brnaba-aws]

Use case

Today in the classifier the model being used by default is using Claude Sonnet 3.5 on Amazon Bedrock.
We need another demo with a fine tuned LLM to showcase the implementation details.

Solution/User Experience

With a fine tuned LLM, the prompt can be reduced (hence number of tokens) and would have an impact on cost and latency.
This demo will also showcase how to build a custom classifier with a model that is fine tuned on specific domain.

Alternative solutions

No alternative solution for now.