Strong consistency is not guaranteed

### [REQUIRED] Describe your environment

* Operating System version: WSL Ubuntu 20.04 on Windows 20H2
* Browser version: Chrome 97.0.4692.71
* Firebase SDK version: master branch (4bd3a888353c26854142b1861f12e0222151af96)
* Firebase Product: firestore

### [REQUIRED] Describe the problem
When I call DocumentReference.get() asynchronously, update the document synchronously, and then do DocumentReference.get() again, I expect to get the last updated document, however, sometimes the old data may be retrieved. (See the code below.)

DocumentReference.get() looks like it's internally treating the request as a query and passing it to the EventManager, but
EventManagerImpl manages the same query in one place, so this probably causes the new request to be lumped in with the old one.

https://github.com/firebase/firebase-js-sdk/blob/4bd3a888353c26854142b1861f12e0222151af96/packages/firestore/src/core/event_manager.ts#L64

#### Steps to reproduce:

The code below (Occasional failures)

#### Relevant Code:
```typescript
it('get document while online with default get options keep strong consistency', () => {
const initialData = { key: 'initial' };
const updatedData = { key: 'updated' };
return withTestDocAndInitialData(persistence, initialData, async docRef => {
// 1. async get
void getDoc(docRef);

// 2. sync update (Or the same with calling functions that update data)
await updateDoc(docRef, updatedData);

// 3. get
// Maybe if request 1 is resolved late, the same data as request 1 will be returned
// I think this behavior is breaking strong consistency.
const dataGotAfterUpdate = await getDoc(docRef);

expect(dataGotAfterUpdate.data()).to.deep.equal(updatedData);
});
});
```