Authentication error in pipeline after batch endpoint invocation

### Operating System

Linux 23.06.30

### Compute Cluster
Standard_NC6s_v3 (Managed identity: User assigned identity)

### Version Information

Python Version: 3.10.11
azure-ai-ml package version: 1.8.0

### Steps to reproduce

1. Run the [custom-output-batch.ipynb](https://github.com/Azure/azureml-examples/blob/main/sdk/python/endpoints/batch/deploy-models/custom-outputs-parquet/custom-output-batch.ipynb) until step 4.6.2

### Expected behaviour

The pipeline job completes without problems.

### Actual behaviour

The prepared_data component fails directly with the following error:
**_Datastore has no credentials, and workspace msi failed to authenticate storage account. Please grant Storage Account Contributor role to workspace msi._**

### Addition information


Start by running custom-output-batch.ipynb through step 4.6.2 with the listed Python, azure-ai-ml, cluster, and managed identity settings. Investigate the prepared_data authentication failure and verify that the pipeline completes without the datastore credential error.

Written by the indexing model from the issue text.