Adds support for skipToken in options for useSuspenseQuery and useBackgroundQuery and soft deprecates skip option

[jerelmiller]

Closes #11093

Adds support for a new skipToken sentinel that can be passed as options to useSuspenseQuery and useBackgroundQuery. This works in place of skip for both of these hooks. As such, the skip option is now labeled as deprecated to nudge people to use this feature, which is generally more type-safe than the skip option.

Breaking change

This PR also changes the way queryRef is returned when skip is true or skipToken is used with useBackgroundQuery. Previously, a queryRef would always be returned, regardless of whether query execution was skipped. This has been updated to conditionally return a queryRef only when query execution is enabled. undefined is returned when initializing the hook skipped.

Before

function Parent() {
  const [queryRef] = useBackgroundQuery(query, skip ? skipToken : undefined);
  //      ^? QueryReference<TData | undefined>

  return <Child queryRef={queryRef} />
}

function Child({ queryRef }: { queryRef: QueryReference<TData | undefined> }) {
  const { data } = useReadQuery(queryRef);
}

After

function Parent() {
  const [queryRef] = useBackgroundQuery(query, skip ? skipToken : undefined);
  //      ^? QueryReference<TData> | undefined

  return queryRef ? <Child queryRef={queryRef} /> : null;
}

function Child({ queryRef }: { queryRef: QueryReference<TData> }) {
  const { data } = useReadQuery(queryRef);
}

Checklist:

• If this PR contains changes to the library itself (not necessary for e.g. docs updates), please include a changeset (see CONTRIBUTING.md)
• If this PR is a new feature, please reference an issue where a consensus about the design was reached (not necessary for small changes)
• Make sure all of the significant new logic is covered by tests

[jerelmiller]

I happened to notice we were importing this function from the wrong place. This has been updated to ensure its using the wrapped version.

[jerelmiller]

There was no reason to memo this return tuple since the tuple itself didn't need to be referentially stable, just the items in the tuple. Typical usage of this hook will destructure this tuple, so no need to memoize it.

[phryneas]

I'd love some minor changes in tests, but generally this is good to go!

[phryneas]

Independently of skip - why would another network request happen here in the first place? Do we need to change variables to provoke that without skip a request would be provoked here?

[jerelmiller]

It's more that I wanted to prove that switching back to skip: true didn't itself have any adverse effects such as tangling it up with some other value, or that it wasn't just a "skips the first time it's true" type of thing. skip is a suspenseful option, so it was just an extra check to ensure I wasn't suspending more than necessary.

You're right though, switching variables is probably a more solid way to show off that it does indeed skip. Let me see if I can make this change to make it more obvious what this test is trying to accomplish.