chore: loose test_interval_task_performance a bit to prevent flaky behaviour in CI

Author: danfimov

tests/scheduler/test_interval_performance.py

@@ -22,14 +22,19 @@ def performance_task(value: int) -> int:
 
         # Wait for 5 executions
         for _ in range(5):
-            msg = await asyncio.wait_for(broker.queue.get(), 2)
+            msg = await asyncio.wait_for(broker.queue.get(), 3)
             execution_times.append(time.time())
             assert msg is not None
 
-        # Check intervals between executions
-        for i in range(1, len(execution_times)):
-            interval = execution_times[i] - execution_times[i - 1]
-            assert 0.95 <= interval <= 2  # Allow some tolerance
+        intervals = [
+            execution_times[i] - execution_times[i - 1]
+            for i in range(1, len(execution_times))
+        ]
+        for interval in intervals:
+            assert 0.5 <= interval <= 3  # Loose bound to catch gross regressions
+
+        average_interval = sum(intervals) / len(intervals)
+        assert 0.8 <= average_interval <= 1.5
 
     finally:
         scheduler_task.cancel()