include api_base in config parameters REST API

[minrk]

this was the missing piece in jupyterlab/jupyter-ai#1478 to get api_base into the config UI

get_supported_openai_params doesn't include client parameters like api_base, so api_base is never included from the rest api.

[srdas]

@minrk The code looks good. Thanks for making this update as it will help users in v3, as requested here: jupyterlab/jupyter-ai#1010 (comment)

I have also tested this with the Ollama GPT-OSS as follows:

1. Set the default port for Ollama to 10000, now that we can set the parameter api_base as required and try the model, which works:

2. Next, change the port to 11434 (default) and see that it fails as Ollama is still running on port 10000 (there is no response on the chat interface, but we can see the log below):

4. The restart Ollama from its default port and retry:

It works again. So the api_base updates are working correctly.

[ellisonbg]

Thanks Min!

[dlqqq]

@minrk Thank you for fixing this for us so quickly! Changes look great, thank you @srdas for testing this locally also.

Merging and releasing now 🎉