Avoid wrapping prediction dataloader twice on TPU

[Liyang90]

Avoid wrapping dataloader with MpDeviceLoader more than once. Add batch_sampler back to the dataloader.

What does this PR do?

Fixes #16572

When doing prediction on TPU with:

trainer=Trainer(accelerator='tpu', devices=8)
trainer.predict(model=model, dataloaders=dm)

the code would crash because the prediction dataloader is wrapped twice with MpDeviceLoader, once in DataConnector._reset_eval_dataloader() and once in PredictionLoop.advance().

To avoid double wrapping, a checking is added in TPUSpawnStrategy.process_dataloader().

The batch_sampler is also added to the wrapped dataloader to avoid warning in prediction_epoch_loop.

Does your PR introduce any breaking changes? If yes, please list them.

None

Before submitting

• Was this discussed/approved via a GitHub issue? (not for typos and docs)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure your PR does only one thing, instead of bundling different changes together?
• Did you make sure to update the documentation with your changes? (if necessary)
• Did you write any new necessary tests? (not for typos and docs)
• Did you verify new and existing tests pass locally with your changes?
• Did you list all the breaking changes introduced by this pull request?
• Did you update the CHANGELOG? (not for typos, docs, test updates, or minor internal changes/refactors)

PR review

Anyone in the community is welcome to review the PR.
Before you start reviewing, make sure you have read the review guidelines. In short, see the following bullet-list:

• Is this pull request ready for review? (if not, please submit in draft mode)
• Check that all items from Before submitting are resolved
• Make sure the title is self-explanatory and the description concisely explains the PR
• Add labels and milestones (and optionally projects) to the PR so it can be classified

Did you have fun?

Yes
Make sure you had fun coding 🙃

[carmocca]

CI status: blocked by #16613