Add the possibility to use Azure AI Foundry model deployments in FXB

[adner]

Azure AI Foundry should be added as a provider that can be selected in the FBX AI settings. This would allow a user to create model deployments in Foundry and use these in FXB. See here for info on how to use the Azure.AI.Inference lib to call models in AAI Foundry.

Of special interest in this documentation where it is discussed how to use the Azure.AI.OpenAI library to call OpenAI deployments in Foundry - this is of interest because of the new GPT-oss model from OpenAI that was just released, and that is probably were useful (and cheap) for FXB. Azure.AI.OpenAI is build on the OpenAI package that is already a reference in FXB.

[adner]

@rappen Is this obsolete, due to the introduction of the FXB "free AI tier"? Or do we still want this issue if we want to add the possibility in the future to add your own Foundry model deployments?

[rappen]

It might be good to support it in the future, so I guess we should procrastinate for a while to see how users would like to proceed.