Not FUD, these are good questions. In my opinion, since you asked, writing an http server using Loom (preview mode and all) is the only reasonable way to get ahead of the market, so to speak. It's not like I am an insider or have any special access, nor do I have any personal proofs to my statements, but I do believe that Project Loom / virtual threads will hold up well in production. I think there will be lots of interesting improvements and use-cases that will be drawn out of loom, and using it now in preview mode would be a reasonable way to get ahead.
That being said, yes everything is very "experimental" right now. But a start-from-scratch no-dependencies project such as yours (or anyone else's for that matter) might be the best approach versus trying to retrofit an existing server.
That being said, yes everything is very "experimental" right now. But a start-from-scratch no-dependencies project such as yours (or anyone else's for that matter) might be the best approach versus trying to retrofit an existing server.