sample_workflow_multi_agent.py fails

- **Package Name**: azure-ai-projects
- **Package Version**: 2.0.0b3
- **Operating System**: Windows
- **Python Version**: 3.12.10

**Describe the bug**
When running the sample, the output indicates the response failed:
`Event 10 type 'response.failed'`

**To Reproduce**
Steps to reproduce the behavior:
1. run `pip install "azure-ai-projects>=2.0.0b1" python-dotenv aiohttp`
2. Run the [sample_workflow_multi_agent](https://github.com/Azure/azure-sdk-for-python/blob/5152e360493c3752d1b5068a512da2ab13157cbe/sdk/ai/azure-ai-projects/samples/agents/sample_workflow_multi_agent.py) code sample
```
import os
from dotenv import load_dotenv

from azure.identity import DefaultAzureCredential
from azure.ai.projects import AIProjectClient
from azure.ai.projects.models import (
PromptAgentDefinition,
WorkflowAgentDefinition,
ItemType,
)

load_dotenv()

endpoint = os.environ["AZURE_AI_PROJECT_ENDPOINT"]

with (
DefaultAzureCredential() as credential,
AIProjectClient(endpoint=endpoint, credential=credential) as project_client,
project_client.get_openai_client() as openai_client,
):
# Create Teacher Agent
teacher_agent = project_client.agents.create_version(
agent_name="teacher-agent",
definition=PromptAgentDefinition(
model=os.environ["AZURE_AI_MODEL_DEPLOYMENT_NAME"],
instructions="""You are a teacher that create pre-school math question for student and check answer.
If the answer is correct, you stop the conversation by saying [COMPLETE].
If the answer is wrong, you ask student to fix it.""",
),
)
print(f"Agent created (id: {teacher_agent.id}, name: {teacher_agent.name}, version: {teacher_agent.version})")

# Create Student Agent
student_agent = project_client.agents.create_version(
agent_name="student-agent",
definition=PromptAgentDefinition(
model=os.environ["AZURE_AI_MODEL_DEPLOYMENT_NAME"],
instructions="""You are a student who answers questions from the teacher.
When the teacher gives you a question, you answer it.""",
),
)
print(f"Agent created (id: {student_agent.id}, name: {student_agent.name}, version: {student_agent.version})")

# Create Multi-Agent Workflow
workflow_yaml = f"""
kind: workflow
trigger:
kind: OnConversationStart
id: my_workflow
actions:
- kind: SetVariable
id: set_variable_input_task
variable: Local.LatestMessage
value: "=UserMessage(System.LastMessageText)"

- kind: CreateConversation
id: create_student_conversation
conversationId: Local.StudentConversationId

- kind: CreateConversation
id: create_teacher_conversation
conversationId: Local.TeacherConversationId

- kind: InvokeAzureAgent
id: student_agent
description: The student node
conversationId: "=Local.StudentConversationId"
agent:
name: {student_agent.name}
input:
messages: "=Local.LatestMessage"
output:
messages: Local.LatestMessage

- kind: InvokeAzureAgent
id: teacher_agent
description: The teacher node
conversationId: "=Local.TeacherConversationId"
agent:
name: {teacher_agent.name}
input:
messages: "=Local.LatestMessage"
output:
messages: Local.LatestMessage

- kind: SetVariable
id: set_variable_turncount
variable: Local.TurnCount
value: "=Local.TurnCount + 1"

- kind: ConditionGroup
id: completion_check
conditions:
- condition: '=!IsBlank(Find("[COMPLETE]", Upper(Last(Local.LatestMessage).Text)))'
id: check_done
actions:
- kind: EndConversation
id: end_workflow

- condition: "=Local.TurnCount >= 4"
id: check_turn_count_exceeded
actions:
- kind: SendActivity
id: send_activity_tired
activity: "Let's try again later...I am tired."

elseActions:
- kind: GotoAction
id: goto_student_agent
actionId: student_agent
"""

workflow = project_client.agents.create_version(
agent_name="student-teacher-workflow",
definition=WorkflowAgentDefinition(workflow=workflow_yaml),
)

print(f"Agent created (id: {workflow.id}, name: {workflow.name}, version: {workflow.version})")

conversation = openai_client.conversations.create()
print(f"Created conversation (id: {conversation.id})")

stream = openai_client.responses.create(
conversation=conversation.id,
extra_body={"agent": {"name": workflow.name, "type": "agent_reference"}},
input="1 + 1 = ?",
stream=True,
metadata={"x-ms-debug-mode-enabled": "1"},
)

for event in stream:
print(f"Event {event.sequence_number} type '{event.type}'", end="")
if (
event.type == "response.output_item.added" or event.type == "response.output_item.done"
) and event.item.type == ItemType.WORKFLOW_ACTION:
print(
f": item action ID '{event.item.action_id}' is '{event.item.status}' (previous action ID: '{event.item.previous_action_id}')",
end="",
)
print("", flush=True)

openai_client.conversations.delete(conversation_id=conversation.id)
print("Conversation deleted")

project_client.agents.delete_version(agent_name=workflow.name, agent_version=workflow.version)
print("Workflow deleted")

project_client.agents.delete_version(agent_name=student_agent.name, agent_version=student_agent.version)
print("Student Agent deleted")

project_client.agents.delete_version(agent_name=teacher_agent.name, agent_version=teacher_agent.version)
print("Teacher Agent deleted")
```
3. Observe the failed response in the output:
```
...
Event 9 type 'response.output_item.added': item action ID 'student_agent' is 'in_progress' (previous action ID: 'create_teacher_conversation')
Event 10 type 'response.failed'
Conversation deleted
Workflow deleted
...
```
**Expected behavior**
The response should succeed