Performance discipline

TL;DR. No speculative optimization, but three things we do measure: single-pane yes-flood throughput against tmux, multi-pane fanout, and reattach latency for big scrollback. Benches live in benches/ per crate via criterion; the release profile uses fat LTO and one codegen unit because shipped binary perf is a goal.

---

We do not optimize speculatively. We do measure:

• Single-pane throughput under a yes flood. tmux is the baseline; we must not be worse.
• Multi-pane fanout: one server, N clients, M panes.
• Reattach latency for sessions with large scrollback.

Benchmarks live in benches/ per crate, using criterion (added when there is code to benchmark). The release profile uses fat LTO and a single codegen unit because final binary perf is a goal.