Treat zombie workers as terminated in tests

Why: Gitea containers may leave SIGKILLed children as zombies until PID 1 reaps them, while kill -0 still reports success.

Evidence: Both timeout tests and the full AgentExecControllerTests suite pass; git diff --check passes.

Impact: Changes test liveness detection only; genuinely running child processes still fail the assertions.

Refs: #567

tests/test_agent_exec.py

@@ -41,6 +41,25 @@ def _canonical_snapshot(repo: Path) -> dict[str, str]:
     return {"head": head, "tree": tree, "status": status, "index": index}
 
 
+def _pid_alive(pid: int) -> bool:
+    """Return true only for a running, non-zombie process."""
+    try:
+        text = Path(f"/proc/{pid}/stat").read_text(
+            encoding="utf-8", errors="replace"
+        )
+    except FileNotFoundError:
+        return False
+    rpar = text.rfind(")")
+    if rpar != -1 and rpar + 2 < len(text) and text[rpar + 2] == "Z":
+        return False
+    return (
+        subprocess.run(
+            ["kill", "-0", str(pid)], check=False, capture_output=True
+        ).returncode
+        == 0
+    )
+
+
 def _write_config(
     repo: Path,
     *,
@@ -740,13 +759,8 @@ class AgentExecControllerTests(unittest.TestCase):
         self.assertNotIn("idle timeout", data["reason"])
         if child_pid_file.exists():
             child = int(child_pid_file.read_text(encoding="utf-8").strip())
-            # Child must not still be alive.
+            # A zombie is terminated even if a container PID 1 has not reaped it.
-            still = subprocess.run(
+            self.assertFalse(_pid_alive(child))
-                ["kill", "-0", str(child)],
-                check=False,
-                capture_output=True,
-            )
-            self.assertNotEqual(still.returncode, 0)
         self._assert_canonical_untouched()
 
     def test_idle_timeout_kills_silent_worker_before_total(self) -> None:
@@ -821,12 +835,8 @@ class AgentExecControllerTests(unittest.TestCase):
         self.assertEqual(cleanup["reason"], "idle timeout")
         self.assertTrue(child_pid_file.is_file())
         child = int(child_pid_file.read_text(encoding="utf-8").strip())
-        still = subprocess.run(
+        # A zombie is terminated even if a container PID 1 has not reaped it.
-            ["kill", "-0", str(child)],
+        self.assertFalse(_pid_alive(child))
-            check=False,
-            capture_output=True,
-        )
-        self.assertNotEqual(still.returncode, 0)
         self._assert_canonical_untouched()
 
     def test_progress_keeps_worker_alive_past_one_idle_interval(self) -> None: