Add staffing timing to the integration harness and commit the verified
run: small train 171ms, forecast train 3.6s, SSE propagation 15ms,
120/120 clients for 300s with 0 drops and p95 5.6ms, 20 heartbeats at
15s, FastAPI-down fallback from the snapshot, and recovery in 1.6s.
Add scripts/integration.sh, which starts the real FastAPI and Go API,
trains small and forecast through POST /api/train, checks SSE scenario
propagation, heartbeat, reconnect, then kills FastAPI to verify snapshot
and last-good forecast fallback before restarting it for recovery.
Add scripts/sse-load, a standard-library load generator that holds 120
concurrent SSE streams for five minutes, posts scenario changes, records
p50/p95/p99 event latency, drops and heartbeat cadence, and exits
non-zero when a contract threshold is missed. Add unit tests, make
targets and README documentation.