Describe the bug
The Cloudreve slave node process crashes with a fatal error: concurrent map read and map write.

To Reproduce
Steps to reproduce the behavior:

1. Set up a Cloudreve environment with at least one Master node and one Slave node.
2. On the Master node, add an offline download task that will be executed by the Slave node. Using multiple simultaneous tasks or a single task with multiple files may increase the likelihood of triggering the bug.
3. Wait for the slave node to complete the download.
4. Observe the slave node's logs (journalctl -u cloudreve.service -f or the log file) at the moment it starts transferring the file to the master. The slave process will crash and exit.

Additional context

• Cloudreve Version: 4.3
• Deployment Architecture: Master-Slave setup (1 Master, 2 Slaves). Slaves communicate with Master through Cloudflare tunnel.
• Host System: The slave node is cloudreve-slave1 running as a systemd service in LXC container.

This appears to be a concurrency issue within the SlaveUploadTask workflow, as indicated by the stack trace. The program exits with status=2/INVALIDARGUMENT.

Journalctl Dump
crash_dump.txt