Add missing word 'bounded' to definition of flatten

[eernstg]

This PR adds the word 'bounded' to the S? case in the definition of flatten. This change is needed for consistency. Here is an example:

typedef Exactly<X> = X Function(X);

extension<X> on X {
  X expectStaticType<Y extends Exactly<X>>() => this;
}

Future<void> test<X extends Future<int>?>(X x) async {
  // ignore: await_only_futures
  var y = await x;
  y.expectStaticType<Exactly<int?>>; // This is what we want.
}

void main() async {
  await test<Future<int>?>(null);
  await test<Future<int>>(Future.value(1));
}

If we do not have the word 'bounded' in said rule then await x will have the type X and the future will not be awaited. The type will be int? as shown above if the rule is changed and 'bounded' is included.

There is no breaking change, and indeed no implementation effort associated with this specification change: the implementations behave as if the spec change had already been performed.

[eernstg]

Hi, Nate!

It is indeed true that some cases in the current implementations behave as if we had that word 'bounded' in the specification already—but it isn't true in all cases. So the situation isn't as simple as it looked at first.

So I'll retract this PR and then we'll have to discuss the topic in more detail in the language team.

Sorry about the bothering you for no good reason!