[ai agents extension] AI agent init fails in westus2 (‘gpt-4o-mini’ not found), but works in westus – enhance error messaging for unsupported locations

[Menghua1]

Description:
When deploying calculator-agent using the azd ai agent and selecting the westus2 region (model is gpt-4o-mini), an error message model 'gpt-4o-mini' not found in AI model catalog appears, as shown in the image below.
However, switching the region to westus works successfully.

Steps to Reproduce:

1. Initialize the project: azd init --template Azure-Samples/ai-foundry-starter-basic.
2. Initialize AI agent: azd ai agent init -m https://github.com/coreai-microsoft/foundry-golden-path/blob/main/idea-to-proto/01-build-agent-in-code/agent-catalog-code-samples/calculator-agent/agent.yaml select the westus2 region

Environment:

• Azd version: azd version 1.21.1-daily.5560360 (commit fa4d348).
• AI Agent Extension version: 0.0.5
• Agent Samples: calculator-agent, learn-french-agent, msft-docs-agent, web-search-agent.

Expected Behavior:
The selected region does not support the model, and initialization will fail immediately. To improve user experience, it is recommended that the CLI prompt the user to select another supported region when the model is unavailable.

JeffreyCA and Rajesh Kamal (@rajeshkamal5050) for notification.

[JeffreyCA]

Currently, if you select a bad/unsupported region, it's not easy to recover from. Re-running azd ai agent init -m results in the same error again:

The user would need to start over from scratch or manually set the AZURE_LOCATION env to a supported region like northcentralus, which isn't apparent from the error.

Another idea may be to prompt for an alternative location if the model couldn't be found in the catalog.

---

Also as part of the composability feature we added special location prompting logic to filter out locations for models that don't have sufficient quota in https://github.com/Azure/azure-dev/pull/4841/changes#diff-1b5e44fe42af1db0e875900c0a49f7a658d598ed67b032c130ad129fe651e864