Resource migration tasks need more graceful handling of resource allocation failure

If a resource migration task fails, it can throw an exception. There may not be a handler for this exception, if this is happening from the task thread, which results in an abort. There's 2 things we should fix here:

1. Ensure there's a handler so this doesn't tear down the entire app.
2. Ensure future accesses to that resource can deal with having a null active resource from allocation failure, to avoid nullptr accesses from just crashing the app later on.

Trace resource migration tasks from the task thread and follow how allocation failures propagate into resource access. Check the failure path for exception handling and the later handling of a null active resource. Done means migration failure no longer aborts the application and subsequent accesses avoid null-pointer crashes.