Codemod act -> await act (2/?)

Similar to the rationale for `waitFor` (see facebook#26285), we should always
await the result of an `act` call so that microtasks have a chance to
fire.

This only affects the internal `act` that we use in our repo, for now.
In the public `act` API, we don't yet require this; however, we
effectively will for any update that triggers suspense once `use` lands.
So we likely will start warning in an upcoming minor.

packages/react-dom/src/__tests__/ReactDOMServerSuspense-test.internal.js

@@ -161,7 +161,7 @@ describe('ReactDOMServerSuspense', () => {
     expect(divB.tagName).toBe('DIV');
     expect(divB.textContent).toBe('B');
 
-    act(() => {
+    await act(async () => {
       ReactDOMClient.hydrateRoot(parent, example);
     });
 

packages/react-dom/src/__tests__/ReactDOMSuspensePlaceholder-test.js

@@ -153,23 +153,15 @@ describe('ReactDOMSuspensePlaceholder', () => {
         );
       }
 
-      act(() => {
+      await act(async () => {
         ReactDOM.render(<App />, container);
       });
       expect(container.innerHTML).toEqual(
         '<span style="display: none;">Sibling</span><span style=' +
           '"display: none;"></span>Loading...',
       );
 
-      act(() => setIsVisible(true));
-      expect(container.innerHTML).toEqual(
-        '<span style="display: none;">Sibling</span><span style=' +
-          '"display: none;"></span>Loading...',
-      );
-
-      await advanceTimers(500);
-
-      Scheduler.unstable_flushAll();
+      await act(async () => setIsVisible(true));
 
       expect(container.innerHTML).toEqual(
         '<span style="display: inline;">Sibling</span><span style="">Async</span>',