Problem
Right now, on a fresh install, the UX leads user to:
0. Install Jan for the first tijme

1. See empty threads page
2. Go to Hub
3. Download Mistral
4. Go back to threads
5. Start chat
6. Chat default to GPT!
7. It obviously doesn't work because no api key is configured.
8. 
9. How do I use Mistral (the local model I just downloaded)?!
10. I go to system monitor. I see nothing.
11. I have to somehow discover Settings > models > click on breadcrumbs > start model
12. Go back to threads
13. Choose Mistral from Model > dropdown
14. Now I can use Mistral. Yikes

This onboarding experience is not good

Success Criteria
A clear and concise description of what you want to happen.

1. A clear user onboarding journey
2. Maybe land them in the hub to start
3. Always default to a local model (started is best)
4. If its a first model download, then autostart the model for the user!

Additional context
Related ticket about improving the remote model experience #994