fix: correct end-of-init message logic for existing project + existing deployment

[v1212]

Summary

Fixes the end-of-init message incorrectly recommending azd up when the user selected an existing project with an existing deployment (no provisioning needed).

Root cause: len(a.deploymentDetails) == 0 was used to detect whether new provisioning is needed, but existing deployments also populate this slice.

Fix: Replace with an explicit needsProvision boolean set only when a new model deployment is created (deploy-from-catalog path). This aligns init.go with the pattern already used in init_from_code.go.

Truth table

Project	Model config	Needs provision?	Message
Existing	Use existing deployment	No	azd deploy
Existing	Deploy from catalog	Yes	azd up
Existing	Skip	No	azd deploy
New	Deploy from catalog	Yes	azd up
New	Skip	Yes	azd up

Follow-up to #8292 — identified by @wbreza in re-review.

[Copilot]

Pull request overview

Fixes the “next steps” messaging at the end of azd ai agent init so it no longer recommends azd up when the user chose an existing Foundry project with an existing model deployment (i.e., when no new provisioning is required).

Changes:

• Introduces an explicit needsProvision flag on InitAction to represent whether the init flow configured a new model deployment that requires provisioning.
• Updates the end-of-init message logic in init.go to use needsProvision instead of len(deploymentDetails) == 0.
• Sets needsProvision = true when the flow falls through to “deploy new model deployment” logic in getModelDeploymentDetails.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
cli/azd/extensions/azure.ai.agents/internal/cmd/init.go	Switches end-of-init guidance to rely on needsProvision (avoids misclassifying existing deployments as needing azd up).
cli/azd/extensions/azure.ai.agents/internal/cmd/init_models.go	Sets needsProvision when configuring a new model deployment (deploy-from-catalog / deploy-new path).