chore: establish and document performance baseline (pre-modification)

[digital-michael]

Goal

Measure and record input-handling performance on the unmodified fork before any SeamGate changes. This baseline is required by the Project Charter (§15) and must be captured before code changes begin.

Metrics to capture (per §15)

• End-to-end keystroke latency (p50/p95/p99) — local loopback
• Mouse cursor latency (p50/p95/p99)
• CPU utilisation (server + client, idle and under load)
• Memory footprint (server + client, steady state)
• Reconnect time after network drop

Acceptance criteria

• Measurements taken on Fedora 43 / Wayland (after chore: verify local build on Fedora 43 / Wayland (unmodified fork) #5 build verified)
• Results committed to doc/perf-baseline.md on a feature/perf-baseline branch
• PR merged to phase/0 before any code modifications land

Notes

Every subsequent phase gate requires a performance report showing no regression vs this baseline.