feat(frontend): support await-tree dump

[chenzl25]

I hereby agree to the terms of the RisingWave Labs, Inc. Contributor License Agreement.

What's changed and what's your intention?

This is stacked on #26515, which adds await-tree roots for batch tasks on compute nodes. This PR completes the frontend side of the dump path so risectl await-tree dump can show the query driver and local-mode frontend execution as well.

• Create an optional frontend await-tree registry using the existing streaming.async_stack_trace setting; no new configuration is introduced.
• Register the full local query execution future as a Local Query root. Await spans from the local executor, including local exchange waits added by feat(batch): support await-tree dump for batch tasks #26515, are therefore attached to the same tree.
• Register the distributed query scheduler driver as a Distributed Query root.
• Implement MonitorService::stack_trace on frontend nodes and return the active query trees in a new frontend_traces field.
• Have meta enumerate frontend workers, collect their monitor responses through the existing frontend internal RPC address, and merge them into the cluster dump.
• Render the result as a new --- Frontend Traces --- section in risectl await-tree dump.

When async stack tracing is disabled, the frontend monitor service returns an empty trace map.

Testing

• cargo check -p risingwave_pb
• cargo check -p risingwave_frontend -p risingwave_meta -p risingwave_ctl
• cargo check -p risingwave_compute
• cargo clippy -p risingwave_frontend -p risingwave_meta -p risingwave_ctl -p risingwave_compute --lib -- -D warnings
• cargo sort --check --workspace --grouped
• cargo fmt --all
• git diff --check

Checklist

• I have written necessary rustdoc comments.
• I have added necessary unit tests and integration tests.
• I have added test labels as necessary.
• I have added fuzzing tests or opened an issue to track them.
• My PR contains breaking changes.
• My PR changes performance-critical code, so I will run (micro) benchmarks and present the results.
• I have checked the Release Timeline and Currently Supported Versions to determine which release branches I need to cherry-pick this PR into.

Documentation

• My PR needs documentation updates.

Release note

[chenzl25]

• feat(frontend): support await-tree dump #26552 👈 (View in Graphite)
• feat(batch): support await-tree dump for batch tasks #26515
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[wenym1]

Seems that we don't have any newly added instrument logic. Do we already have instrument_await in frontend code, but just not enabled because we didn't register the await tree root in frontend?

[chenzl25]

@wenym1 You are right—there was no existing instrument_await in frontend. I pushed f801fa8afa to add the missing instrumentation:

• query drivers now show wait_stage_event;
• every StageRunner is registered as its own frontend await-tree root ({query_id}/stage/{stage_id}), since the task-local context does not cross the compute-runtime spawn;
• stage scheduling/status/shutdown, frontend root/local executor build and poll, result-channel backpressure, and batch exchange execute/get-data RPCs are instrumented.

The local query root can also reuse existing batch/storage child spans. Thanks for catching this.