Update default HuggingFaceLocalChatGenerator model

[vblagoje]

Is your feature request related to a problem? Please describe.
The defaults model and example models in our HF integrations are outdated. Today (Dec 2025) we have some rather interesting small models that one can still run locally on the of the shelf laptop and get decent enough results. Take for example https://huggingface.co/Qwen/Qwen3-0.6B a Bert size model yet much more capable. However, some of these newer smaller models also have enable_thinking flag that should be elevated to init param.

Describe the solution you'd like
Update the default model HuggingFaceH4/zephyr-7b-beta to be /Qwen/Qwen3-0.6B
Add enable_thinking init param