Open Radar

 
`ModelContext` currently has two methods to fetch a model by ID: `model(for:)` and `registeredModel(for:)`. The latter only returns a model that is already in the context. The former may fetch a model from the model container in addition to the context, but it crashes if the model was not found.

There should be a third method, similar to `NSManagedObjectContext.existingObject(with:)`, that can fetch a model from the model container but does not crash if the model was not found. Or, perhaps `model(for:)` should be updated to not crash.

One use case for such an API would be passing a model ID across concurrency domains. The receiver might use that model ID to fetch the model from the model context/container. However, since this is in a different concurrency domain, the model could have been deleted by the time the receiver tried to fetch the model. This is an expected edge case that should gracefully return `nil` or throw an error rather than crashing.

Example implementation:
```
extension ModelContext {
   func model<T: PersistentModel>(for persistentModelID: PersistentIdentifier) throws -> T? {
      if let model: T = registeredModel(for: persistentModelID) {
         return model
      }

      var fetchDescriptor = FetchDescriptor<T>(
         predicate: #Predicate {
            $0.persistentModelID == persistentModelID
         }
      )
      fetchDescriptor.fetchLimit = 1

      return try fetch(fetchDescriptor).first
   }
}
```