1.7.0.54

Overthrow

A `RequestSpawn` that touches a record whose instance is already in the world (e.g. a self-spawned character) does not fail fast. The engine retries the collision several times a second, forever, never completing and never invoking the callback.

Even on collections guaranteed to have no live stored instance (AIGroup/AIWaypoint), a batch of requests intermittently never invoked its completion callback and blocked every subsequent save for the whole session

## Steps to reproduce

On a world with persistence active and a save containing records for entities that self-spawn on load, call `RequestSpawn` for a record whose instance is already live → observe the error repeating several times per second indefinitely

## Expected

- `RequestSpawn` on an already-instantiated record fails fast with a scripted error result (the id collision is detectable up front), instead of retrying forever.
- A spawn request that cannot complete must not block unrelated save requests; if the system serializes spawn-vs-save, a stuck request needs a timeout.