Integrate Ollama

[bauersimon]

We've gotten many requests for Ollama plus a PR. This would also allow us to run models locally or add new ones that are not yet released.

However, we need to do a few preparations / settle a few questions first before we can do that:

• having multiple providers means we need a mechanic to detect them (cause i.e. people should be able to use ollama without having an openrouter API key)
  • add an Available to a provider that checks if it is available/installed/configered with an API key
• i.e. Ollama is a provider that needs to be installed locally. should we handle the installation if one wants to use it?
  • yes cause we need it for testing in the CI
  • same behavior as symflower, i.e. check if it is installed, and install if not
• i.e. Ollama is a provider that is running as a background service with an API endpoint... should we make sure it runs?
  • yes
• there are some other parameters to an Ollama provider. I.e. the local IP and/or port it is running on. how can a user configure these?
  • for now assume defaults
• the PR Ollama provider #27 crafts the API requests manually, but Ollama has experimental support for the OpenAI API that we already use with openrouter: https://github.com/ollama/ollama/blob/main/docs/openai.md so use that instead

[zimmski]

Moved rest of tasks to #100