Fix #319: record event runtime for events acked via stream.take()

[wbarnha]

What

When an agent consumes a stream with stream.take(max_, within=...), the event runtime metric (events_runtime, surfaced e.g. as Prometheus events_runtime_ms_sum) stayed at zero.

Fixes #319.

Why

take() disables acking (enable_acks = False) and acks its buffered events manually in its finally, long after the main iteration loop yielded them. The main loop only calls on_stream_event_out when do_ack is set, so for the take() path the out-event is delivered by Stream.ack() instead.

But Stream.ack() called on_stream_event_out(tp, offset, self, event) without the sensor state. Monitor.on_stream_event_out needs that state (which carries time_in) to compute time_total; with state=None it does nothing, so no runtime was ever recorded for taken events.

How

• SensorDelegate.on_stream_event_in now stashes the per-sensor state dict on the event itself (event.sensor_state). Both the Python (_py_aiter) and Cython (_cython/streams) iterators call this same delegate method, so the state is captured regardless of which iterator runs — no .pyx recompile needed.
• Stream.ack() reads event.sensor_state, forwards it to on_stream_event_out, then clears it so a second ack can't double-count.
• Event / EventT gain a sensor_state attribute (initialised to None, added to EventT.__slots__).

The normal (acks-enabled) path is unchanged: the main loop still fires on_stream_event_out with its local state and never routes through Stream.ack().

Test

Added test_take__records_event_runtime in tests/functional/test_streams.py: attaches a Monitor, sends a value, consumes one buffered batch via take(), and asserts monitor.events_runtime is non-empty. Verified it fails on master (assert deque([])) and passes with the fix, under both the Cython and NO_CYTHON stream iterators. Updated the Stream.ack unit test for the new forwarded/cleared sensor_state argument.

🤖 Generated with Claude Code

---

Generated by Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.15%. Comparing base (3073eb9) to head (8224172).

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #701      +/-   ##
==========================================
+ Coverage   94.14%   94.15%   +0.01%     
==========================================
  Files         104      104              
  Lines       11136    11141       +5     
  Branches     1201     1201              
==========================================
+ Hits        10484    10490       +6     
+ Misses        551      550       -1     
  Partials      101      101              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.