Endpoint deployment error due to principal roles given to unfound resources in the RG

Error when deploying as an endpoint to AI Studio.

Cause: Setting roles for AOAI and AI Search throws the error: Resources not found. The code gives roles based on the connection name, not the resource name.

**Resolution steps:**

1) Add to the provision.yaml:
AZURE_OPENAI_RESOURCE_NAME: "${aoai.aoai_resource_name}" # necessary for the endpoint deployment
AZURE_SEARCH_RESOURCE_NAME: "${search.search_resource_name}" # necessary for the endpoint deployment

This will update the env variables with the name of AOAI and AI Search Resources

2) Modify deploy.py to give roles based on the resource names (not the connection names as originally):

- for AOAI:

scope=f'''/subscriptions/{os.environ["AZURE_SUBSCRIPTION_ID"]}/resourceGroups/{os.environ["AZURE_RESOURCE_GROUP"]}/providers/Microsoft.CognitiveServices/accounts/{os.environ["AZURE_OPENAI_RESOURCE_NAME"]}''',
role_name="Cognitive Services OpenAI User",
principal_id=endpoint.identity.principal_id
)

- for AI Search: scope=f'''/subscriptions/{os.environ["AZURE_SUBSCRIPTION_ID"]}/resourceGroups/{os.environ["AZURE_RESOURCE_GROUP"]}/providers/Microsoft.Search/searchServices/{os.environ["AZURE_SEARCH_RESOURCE_NAME"]}''',
role_name="Search Index Data Contributor",
principal_id=endpoint.identity.principal_id
)

Beginne mit provision.yaml, um die Umgebungsvariablen für die Ressourcennamen von AOAI und AI Search zu prüfen, und überprüfe anschließend deploy.py, wo die Endpunktrollen zugewiesen werden. Stelle den Endpunkt bereit und verifiziere, dass die Rollenscopes die tatsächlichen Ressourcennamen verwenden und kein resources-not-found-Fehler auftritt.

Vom Indexierungsmodell aus dem Issue-Text verfasst.