fix: data race in WeComBot processedMsgs map access

len(c.processedMsgs) was read outside the lock while another
goroutine could be writing to the map. Move the cleanup check
inside the existing critical section.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

pkg/channels/wecom.go

@@ -339,14 +339,11 @@ func (c *WeComBotChannel) processMessage(ctx context.Context, msg WeComBotMessag
 		return
 	}
 	c.processedMsgs[msgID] = true
-	c.msgMu.Unlock()
-
 	// Clean up old messages periodically (keep last 1000)
 	if len(c.processedMsgs) > 1000 {
-		c.msgMu.Lock()
 		c.processedMsgs = make(map[string]bool)
-		c.msgMu.Unlock()
 	}
+	c.msgMu.Unlock()
 
 	senderID := msg.From.UserID