fix: run post execute hooks when stream_result fails before streaming

[digit50]

closes #967

stream_result and astream_result call the pre_run_execute_call hook first, then process the control flow parameters, validate the halt conditions and fast-forward with run(), but only the later try block calls the post hook on an error. When anything in that early section raised (an action failing during the fast-forward, or an unknown halt_after name), post_run_execute_call never ran, so LangfuseBridge left its root span and its Langfuse propagate_attributes context attached. Later spans in the same context landed in the failed call's trace with its session and user ids, and the same app's next run() emitted no run span because the app id stayed registered.

Changes

• In Application.stream_result and Application.astream_result, the section between the pre hook and the existing try is wrapped so that an exception calls call_post / acall_post with that exception and is then re-raised. This matches what a failing run() already does. The rest of the diff in application.py is the indentation that comes with the wrap.
• No change to the Langfuse bridge itself.

How I tested this

• Added test_langfuse_bridge_stream_result_failure_closes_trace (parametrized over a failing fast-forward and an unknown halt_after name) and test_langfuse_bridge_astream_result_failure_closes_trace to tests/integrations/test_burr_langfuse.py. They use the graph from the issue (a -> b, where a raises) and check that the stream_result root span is ended with an error status, that no span is still current, that the Langfuse attribute context has been exited, and that the same app's next run() emits its own run root with no parent. All three fail on main (the stream_result span is never exported) and pass with the change.
• pytest tests/integrations/test_burr_langfuse.py: 12 passed.
• pytest tests/core: 396 passed, 4 failed, identical with and without the change. The 4 failures are test_graphviz_display.py::test_visualize_dot_output, which needs the Graphviz dot executable that is not installed on my machine.
• The reproduction script from the issue now prints the expected output for all three cases.

Notes

The ordering in stream_result predates the Langfuse integration, so any lifecycle hook that pairs pre_run_execute_call with post_run_execute_call (for example the OpenTelemetry bridge) was also missing its post call on these failures.

Checklist

• PR has an informative and human-readable title (this will be pulled into the release notes)
• Changes are limited to a single goal (no scope creep)
• Code passed the pre-commit check & code is left cleaner/nicer than when first encountered.
• Any change in functionality is tested
• New functions are documented (with a description, list of inputs, and expected output)
• Placeholder code is flagged / future TODOs are captured in comments
• Project documentation has been updated if adding/changing functionality.

[elijahbenizzy]

LGTM. Checked that the success-path hook order is identical to main and the post hook fires once; the _run_call_var guard makes the except path safe. Thanks for the parametrized repro. Elijah