[Bug] `MonitoringRecordBuffer._flush_loop` sleeps an extra `flush_interval` after stop signal

sdk/nexent/monitor/monitoring.py:1081-1099:

def _flush_loop(self) -> None:
    while self._running:
        try:
            now = time.time()
            ...
        except Exception as e:
            logger.error(f"Error in monitoring flush loop: {e}")

        for _ in range(10):
            if not self._running:
                return
            time.sleep(self._flush_interval / 10)

Two issues:

1. Latency after exception. If _flush_to_db raises, the loop logs and falls through into the inner sleep. There is no fast-exit on error, so a transient DB outage means a full flush_interval lag on every error, even if the underlying problem cleared within a second. With the default 30 s flush interval that's 30 s of backed-up buffer.

2. Shutdown can hang up to flush_interval / 10 seconds. During the currently-active time.sleep slice, the if not self._running check has already passed. With flush_interval = 30 (default), that's a 3 s shutdown delay. Set flush_interval = 300 (5 min) for low-volume tenants and shutdown blocks 30 s.

A threading.Event solves both:

def __init__(self):
    ...
    self._stop_event = threading.Event()

def _flush_loop(self):
    while not self._stop_event.is_set():
        try:
            ...
        except Exception as e:
            logger.error(...)
        self._stop_event.wait(timeout=self._flush_interval)

def stop(self):
    self._stop_event.set()

Event.wait(timeout=...) returns immediately when set() is called from another thread.

Category: A/H. Severity: Low.

Start in sdk/nexent/monitor/monitoring.py:1081-1099 at MonitoringRecordBuffer._flush_loop, then inspect its init and stop() methods. Trace both the exception path and shutdown path, and verify that the stop signal interrupts waiting immediately while errors do not add an unnecessary full interval delay.