test: regression tests for #5282 (after_run_callback on BaseNode roots)

[surfai]

Summary

Regression test for #5282: Runner._run_node_async does not dispatch plugin_manager.run_after_run_callback on the BaseNode path (the runners.py:427 TODO).

• Baseline test — confirms on_user_message_callback, before_run_callback, and on_event_callback all fire on a Workflow root (they do, via _consume_event_queue at runners.py:619).
• Regression anchor — strict xfail asserting after_run_callback == 1. Passes as xfail on stock 2.0.0a3. When the TODO is wired, delete the @xfail decorator and it flips green.
• Workaround proof — WorkaroundRunner subclass wrapping run_async to dispatch run_after_run_callback post-drain. Confirms the fix shape is a single await ic.plugin_manager.run_after_run_callback(invocation_context=ic) after the event queue drains.

The fix itself (one dispatch call after the _consume_event_queue loop in _run_node_async) is left to the maintainer — this PR provides the test harness and regression guard.

ContextVar note for WorkaroundRunner adopters

The WorkaroundRunner in the test uses self._last_ic to stash the InvocationContext. This is safe for single-invocation tests but races under concurrent run_async calls on a shared Runner instance — the second call overwrites _last_ic before the first call's post-drain dispatch. Production use of this pattern should store ic on a contextvars.ContextVar scoped to the async task instead of self.

adk api_server injection gap

The WorkaroundRunner subclass is only viable for teams instantiating Runner directly. adk api_server hardcodes Runner(...) at adk_web_server.py:737 with no injection point for a custom subclass. Teams using adk api_server (or adk web) have no workaround short of monkey-patching or the in-graph terminal-FunctionNode pattern described in #5282. Worth considering when the runners.py:427 fix lands — ensuring _run_node_async dispatches natively covers both paths.

Test plan

• test_workflow_root_dispatches_pre_run_and_event_hooks — PASSED on 2.0.0a3
• test_workflow_root_after_run_callback_not_dispatched — XFAIL (strict) on 2.0.0a3
• test_workaround_runner_dispatches_after_run_callback — PASSED on 2.0.0a3

When runners.py:427 is resolved:

• Remove @pytest.mark.xfail from test_workflow_root_after_run_callback_not_dispatched — it should PASS

Fixes #5282