MHD tasks are now run in worker#0, which is already used by many other modules like logging and maxadmin. That make API server unstable.

My scenario is, somehow the disk crashed, and the logging task just hung when writing file to disk, that made the worker#0 stopped working. I wanted to restart service by calling API interface, but worker#0 did not respond to any command. If MHD tasks run in an independent thread, at least I got a chance to rescue.

As API server is a vital part of DevOps, will you guys consider to run MHD tasks in an independent thread？