Support specification of llm model in Bee UI

[planetf1]

Related Slack thread

Related issues

• Link any previous issues that might be related here.

Description

Allow LLM Model used by agent to be modified via UI when running the bee stack

Motivation

An agent created via the Bee UI will always use the default LLM for the provider configured via LLM_BACKEND

A new agent can be created via the API. This is documented in the bee stack readme ie:

➜  bee-stack git:(main) curl -X POST \
  "${BEE_API:-localhost:4000}/v1/assistants" \
  -H "Authorization: Bearer ${BEE_API_KEY:-sk-proj-testkey}" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Granite assistant",
    "model": "granite3.1-dense:8b"
  }'

However I think this would be more user friendly if available through the UI, especially as different agents will work better/worse with different LLMs, so this facilitates experimentation.

Ideas

1. Consider supporting OLLAMA_MODEL or similar to specify the default model used by Bee when creating an agent
2. Further, Allow the Model to be updated in the Bee UI

Open Questions

Additional context

Add any other context or screenshots here.