Refactor distributed_sampler update in data_connector.py for distributed strategy

[ninginthecloud]

Proposed refactor

This issue follows @ananthsub 's #11756 to move strategy-specific dataloader logic to the stategies.

Motivation

_prepare_dataloader() function in data_connector.py handles the logic whether the dataloader needs to update the distributed sampler when user is using distributed strategy (see the following code link)
https://github.com/PyTorchLightning/pytorch-lightning/blob/cc43d07db1ab77385feff04c01f040c5cad805a9/pytorch_lightning/trainer/connectors/data_connector.py#L360-L372

However, modifying sampler in dataloader can be pushed directly to strategy class. Similar to issue #12216 , we can move the logic to Strategy base class. For any strategy classes have is_distributed=True flag, they can verify if the dataloader need to update the sampler.

Pitch

We can move the logic to Strategy base class. For any strategy classes have is_distributed=True flag, they can verify if the dataloader need to update the sampler.

Additional context

cc: @edward-io @four4fish @ananthsub

---

If you enjoy Lightning, check out our other projects! ⚡

• Metrics: Machine learning metrics for distributed, scalable PyTorch applications.

• Lite: enables pure PyTorch users to scale their existing code on any kind of device while retaining full control over their own loops and optimization logic.

• Flash: The fastest way to get a Lightning baseline! A collection of tasks for fast prototyping, baselining, fine-tuning, and solving problems with deep learning.

• Bolts: Pretrained SOTA Deep Learning models, callbacks, and more for research and production with PyTorch Lightning and PyTorch.

• Lightning Transformers: Flexible interface for high-performance research using SOTA Transformers leveraging Pytorch Lightning, Transformers, and Hydra.

cc @justusschock @awaelchli @rohitgr7 @ninginthecloud @otaj @tchaton @akihironitta @carmocca