Normally, any new at-exit functions are added to the parent. This means they are not called when the seEval is done but rather when the whole context is cleaned up. If you turn on this flag, at-exit functions created inside the eval are called when the eval is finished.
This may seem like a good idea, but there is an important caveat. At-exit functions normally clean up resources. A call to seEval will return a value to you, and that value may be dependent on those resources. If the at-exit functions have been called, the value is using resources that have been cleaned up. This is why you usually want all at-exit functions held until the context is being destroyed, so you know all such values are no longer used.