Not auto add DistributedSampler for DDP training

[bobofzhang]

🐛 Bug

in 0.72, even if we don't set sampler, pytorch_lightning will not add DistributedSampler for us.

To Reproduce

the reason is in pytorch, if we don't set sampler, pytorch will add a sampler for us.
in pytorch's dataloader.py:

         if sampler is None:  # give default samplers
            if self._dataset_kind == _DatasetKind.Iterable:
                # See NOTE [ Custom Samplers and IterableDataset ]
                sampler = _InfiniteConstantSampler()
            else:  # map-style
                if shuffle:
                    sampler = RandomSampler(dataset)
                else:
                    sampler = SequentialSampler(dataset)

but in pytorch_lightning we check whether sampler is None to decide to add sampler
in data_loading.py funciton auto_add_sampler:

        no_sampler_added = dataloader.sampler is None

because pytorch have default sampler for us, which is not None, pytorch_lighting will not automatically add sampler.