Add Ionos AI Model Hub provider

[muety]

The Ionos AI Model Hub provides LLM inference as a service, billed on a per-token basis, similar to OVHCloud (#5066), Scaleway (#3288) and others. They currently support different Llama- and Mistral variants as well as image- or embedding models and expose an OpenAI-compatible API.

Documentation: https://docs.ionos.com/cloud/ai/ai-model-hub.

I also wrote a short blog article recently on using that service with JetBrains and VSCode, which might be helpful during implementation.