Feature Request: Add fallback_to_parent property to Agent

[takashikik]

Is your feature request related to a problem? Please describe.
Currently, the Agent has a disallow_transfer_to_parent setting to prevent transferring control to the parent agent. However, there are scenarios where the opposite behavior is desired: forcing the agent to always transfer control back to its parent. Without this feature, it's difficult to ensure that a child agent reliably returns to the parent's context after completing its task.

Describe the solution you'd like
I propose adding a new boolean property to the Agent, named fallback_to_parent.

When fallback_to_parent is set to True, the agent must transfer control to its parent agent after its execution is complete. This would act as the inverse of disallow_transfer_to_parent.

Describe alternatives you've considered
One could manually implement a transfer to the parent at the end of every tool within the agent. However, this approach is repetitive and error-prone. A dedicated property on the agent itself would provide a cleaner, more declarative, and more reliable way to manage the conversation flow.

Additional context
This feature would be particularly useful for creating hierarchical agent structures. For example, a parent agent could delegate a specific, self-contained task to a child agent, and with this new property, it can be guaranteed that the conversation flow returns to the parent to continue the main process. This provides more robust control over agent routing.