fix(log): make console sink non-blocking to avoid event-loop stalls

[biao-1010]

Problem

The console loguru sink writes to sys.stdout synchronously on the event loop
thread. The file sink and the trace sink both pass enqueue=True, but the
console sink does not:

# astrbot/core/log.py  (_setup_loguru)
cls._console_sink_id = _loguru.add(
    sys.stdout,
    level="DEBUG",
    colorize=True,
    # <-- missing enqueue=True
    ...
)

When the console blocks (Windows QuickEdit with text selected, a stalled
terminal, a full pipe), write() blocks, and with it the whole asyncio event
loop.

Evidence

On a Windows deployment, logs/event_loop_watchdog.log captured thread stacks
showing the main thread stuck at:

astrbot/core/log.py:140  emit
  -> loguru._handler.emit
  -> loguru._simple_sinks.py:16  write
    -> self._stream.write(message)

for 32.938s and 34.828s. The lag monitor separately reported
Event loop lag detected: 201.203s (threshold 15.000s).

During that window the platform websocket heartbeat could not be sent
(ClientConnectionResetError: Cannot write to closing transport), so the bot
lost its connection for roughly five minutes.

Fix

Add enqueue=True to both console sink registrations, so the console sink
behaves like the file and trace sinks. Logging then goes through loguru's
background queue and a blocking console can no longer stall the event loop.

Notes

• Output format, level and colour are unchanged.
• Same trade-off as the existing file sink: on a hard crash the very last
  queued lines may be lost.
• Tested on Windows 11 with AstrBot 4.28.1: restart clean, logs still written
  to file, no errors.

Summary by Sourcery

Bug Fixes:

• Prevent blocking console writes from stalling the asyncio event loop and disrupting long-lived websocket connections.

[sourcery-ai]

Hey - I've found 1 issue

Prompt for AI Agents

Please address the comments from this code review:

## Individual Comments

### Comment 1
<location path="astrbot/core/log.py" line_range="517" />
<code_context>
                         sys.stdout,
                         level=configured_level,
                         colorize=True,
+                        enqueue=True,  # NOTE(n100): non-blocking, see above
                         filter=lambda record: (
                             not record["extra"].get("is_trace", False)
</code_context>
<issue_to_address>
**issue (bug_risk):** When `configure_logger()` replaces the console sink while its background worker is blocked in `sys.stdout.write()`, `_loguru.remove(cls._console_sink_id)` waits for that queued sink thread to finish. Because `configure_logger()` runs on the event-loop thread, logger reconfiguration still stalls the event loop until the console becomes writable.

**Triggers:** When the log level is reconfigured while the console output is blocked.

**Suggested fix:** Perform sink replacement outside the event-loop thread, or update the sink without synchronously joining a blocked enqueue worker.

```suggestion
                        enqueue=False,  # Avoid joining a blocked worker during reconfiguration
```
</issue_to_address>

Sourcery assessment

Approval pending. 1 finding to address first.

Blocking findings: astrbot/core/log.py:517

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

[sourcery-ai]

issue (bug_risk): When configure_logger() replaces the console sink while its background worker is blocked in sys.stdout.write(), _loguru.remove(cls._console_sink_id) waits for that queued sink thread to finish. Because configure_logger() runs on the event-loop thread, logger reconfiguration still stalls the event loop until the console becomes writable.

Triggers: When the log level is reconfigured while the console output is blocked.

Suggested fix: Perform sink replacement outside the event-loop thread, or update the sink without synchronously joining a blocked enqueue worker.

Suggested change

	enqueue=True, # NOTE(n100): non-blocking, see above
	enqueue=False, # Avoid joining a blocked worker during reconfiguration