Add local Ollama setup and testing scripts for native OpenAI API compatibility

[martin-gaievski]

Added capabilities for setting up model that is natively compatible with OpenAI API and hosted locally. That applies to:

• github CI, added new action that sets up model and executes integration tests that retrieves judgment ratings from LLM
• local development and testing. model is started in local server, and ml-commons model connector can be setup to use it as remote model.

Part of this change are slight improvements in LLM judgment logic, I have added multiple formats of possible response to support basic models.

Resolves issue

#181

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.

[fen-qin]

Really cool to see Ollama can be used for CI tests. Glad to know Ollama has compatibility with OpenAI Chat - OpenAI compatibility and can bridge with OpenAI schema. Great Idea, Martin

[martin-gaievski]

Really cool to see Ollama can be used for CI tests. Glad to know Ollama has compatibility with OpenAI Chat - OpenAI compatibility and can bridge with OpenAI schema. Great Idea, Martin

thanks for feedback and for pointers Fen. I pushed commit that implemented direct connection between ml-commons and ollama for CI and local testing.
I especially like the local testing feature, so we don't need to rely on behemoth models hosted remotely and can use fully local setup for simple tasks