fix: harden OpenSim session readiness

docs/grid-agent-session.md

@@ -37,11 +37,14 @@ server failures retry with exponential backoff. `reconnect.maximum_delay_seconds
 is a hard cap. A server retry hint is a minimum up to that cap. Per-instance
 jitter is bounded by `jitter_basis_points`, preventing synchronized reconnects,
 and a connection that remains up for `stable_reset_seconds` resets the failure
-streak.
+streak. A connected generation that does not reach full event-queue readiness
+within `readiness_timeout_seconds` is logged out and retried instead of
+remaining degraded indefinitely.
 
 Backoff defaults are one second initially, 60 seconds maximum, 20 percent
-jitter, and a 120-second stable reset window. All waits use Tokio timers inside
-`select!` with cancellation/control; there are no blocking sleeps.
+jitter, a 30-second readiness deadline, and a 120-second stable reset window.
+All waits use Tokio timers inside `select!` with cancellation/control; there are
+no blocking sleeps.
 
 ## Generation and work safety
 

docs/live-grid-smoke.md

@@ -40,8 +40,10 @@ With the opt-in and all three `.env` values present, every live case is compiled
 without `#[ignore]`. The test run must report zero ignored cases. Test bodies do
 not print-and-return on missing OpenSim data: a missing inventory item, outfit
 link, appearance update, covenant response, current-region lookup, capability,
-or object packet is a real failure. Login uses the account's OpenSim `last`
-location instead of a hardcoded Second Life region.
+or object packet is a real failure. Live fixtures serialize access to the
+dedicated account even when the Rust test runner uses multiple threads. Login uses the account's OpenSim `last`
+location instead of a hardcoded Second Life region, with one bounded `home`
+fallback when the grid reports that the last region is unavailable.
 
 `GRID_LOGIN_URL` is used directly as the LLSD login endpoint; the harness does
 not substitute a Second Life login host or path. The login request includes the