[Core] Expose fallback_strategy in TaskInfoEntry and ActorTableData

[nadongjun]

Description

Add observability for fallback_strategy in State API and GCS.

While Ray currently provides visibility for label_selector (#53423), there is no mechanism to observe the fallback_strategy from outside the system.

This PR exposes fallback_strategy in TaskInfoEntry and ActorTableData. The ability to read and record fallback_strategy is essential for our custom autoscaler development. When primary label_selector constraints cannot be met, the autoscaler must access these recorded fallback strategies to prioritize and allocate alternative devices.

Beyond autoscaling, adding this feature will provide a better debugging experience by allowing users to transparently track the entire scheduling intent, including the fallback_strategy for both tasks and actors.

Related issues

Related to #51564

Additional information

from ray import serve  
import ray  
from ray.util.scheduling_strategies import NodeLabelSchedulingStrategy, In, Exists  

@serve.deployment(  
    name="soft_docker_deployment",  
    ray_actor_options={  
        "label_selector": {"docker-image": "in(test-image)"},
        "fallback_strategy": [
            {"label_selector": {"docker-image": "in(test-image2)"}},
        ]  
    }
)
class SoftDockerDeployment:  
    def __call__(self, request):  
        node_labels = ray.get_runtime_context().get_node_labels()  
        return {  
            "message": "Hello from soft-docker deployment!",  
            "node_labels": node_labels  
        }  
  
if __name__ == "__main__":  
    serve.start(http_options={"host": "0.0.0.0", "port": 8000})  
    serve.run(SoftDockerDeployment.bind())

GlobalStateAccessor.get_actor_table

ray list actors --detail

[gemini-code-assist]

Code Review

This pull request successfully exposes the fallback_strategy for tasks and actors across Ray's state API and GCS, involving changes to Protobuf definitions, the C++ backend, and the Python state API. The implementation is largely correct and well-aligned with the stated goals. However, I've identified two issues: one in the C++ backend concerning an incorrect check for the fallback_strategy, and another in the Python state API where a field is missing from a dataclass, which would likely cause a runtime error. My review provides specific suggestions to address these points.

[nadongjun]

@edoakes @jjyao Gentle ping. Any thoughts on this?

[edoakes]

LGTM. Thanks for the contribution

Could you add an integration test for the state API to assert that the information is populated correctly?

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

[nadongjun]

@edoakes I've fixed the issue where fallback_strategy was missing during TaskInfo conversion and added both integration and GCS unit tests to verify the fix. Please take another look!

• 9d257d2

[nadongjun]

@edoakes @MengjinYan @israbbani Gentle ping.

[edoakes]

Thanks 👍

[edoakes]

@nadongjun there is a merge conflict

[nadongjun]

@edoakes Resolved the merge conflict.