"There are three kinds of lies: lies, damned lies, and benchmarks."
I believe into user experience. Site is not slow when some benchmark says so, but when users complain for it's speed.
Based on my own experience, I find Misago's speed all right.
Those 30 seconds for first byte look pretty scary but don't worry. Thats how Misago is configured here (for now at least). Misago runs as server. It receives HTTP requests from server, generates responses and returns them to server which in turn sends them back to users browsers. Every 15 minutes killing policy nails all processes on server to free resources. If somebody sends request to Misago after that time, this will cause start of new Misago's server, which takes around 30 seconds to run app.
Keep-alive won't be happening because it makes http server take some control away from app one and this kills performance.