[Bug] Prompted to pick/create Azure resources to use the GitHub model picker when running a Foundry model locally.

### Describe the issue

It's entirely possible I'm doing something wrong here or this isn't supported.

After stepping away from Foundry Local for a while I revisited it tonight. I installed the latest version published to WinGet 10.0.26300. I can download models from the terminal, chat with them, hit the local url like I would expect.

I have the latest of Foundry Toolkit extension 1.6.5 for VS Code VS Code. In the previous version of the toolkit I used I was able to download models and interact with them in the playground but selecting them as a model from the Copilot settings didn't work. But I was able to pick a local model from the list.

Today the playground in VS Code works the same, I can load a model, interact with it, get responses etc. The Copilot settings though I am not able to pick the local models at all. I'm presented with the message is "select a foundry project in foundry toolkit extension to use foundry models".

In the toolkit panel going to Set Foundry Project pops a subscription picker and then it asks for a resource group.

Why do I need to create resources in Azure to interact with a local model? This is a home lab, cost controls are extremely important. If I need to create some objects in Azure to make it work that not an issue. What's unclear to me is why I need to do that in the first place for a local model or what the cost implications might be for that subscription.

### To reproduce

- Install the latest Foundry Toolkit VS Code extension
- From the Foundry Toolkit pane add a model, I used Phi-4mini-reasoning-generic-cpu and qwen2.5-coder-7b-instruct-generic-gpu models this time
- Load any model in the playground, chat with it to see if its working
- In the Copilot model selector settings, look for the running local model under Microsoft Foundry via Foundry Toolkit to use, there are none

### Urgency

Not urgent, this me experimenting in my home lab accounts

### Platform and architecture

Windows X64

### OS Version

Windows 11 25H2

### Installation type

Released package/binary

### Foundry Local version

1.6.5

### API or surface area

CLI

### Hardware acceleration/backend

CPU

### Backend/runtime version

_No response_

Start by reproducing the issue with Foundry Local 1.6.5 and the Foundry Toolkit VS Code extension: load a local model in the playground, then inspect the Copilot model selector and the Set Foundry Project flow. Compare local model discovery with the Azure subscription and resource-group prompts. Done means the expected local-model behavior is restored or the required setup and cost implications are clearly documented.

Written by the indexing model from the issue text.