Move reload_dataloaders_every_n_epochs to the DataHooks class

🚀 Feature

Motivation

We are auditing the Lightning components and APIs to assess opportunities for improvements:

• https://docs.google.com/document/d/1xHU7-iQSpp9KJTjI3As2EM0mfNHHr37WZYpDpwLkivA/edit#
• https://github.com/PyTorchLightning/pytorch-lightning/issues/7740

reload_dataloaders_every_n_epochs is an argument to the Trainer constructor. However, this could be a property of the DataHooks class, instead of the trainer, as whether to initiate reloading the dataloading every n epochs should be determined by the actor providing the dataloaders (e.g. the LightningModule or LightningDataModule).

This is very similar to #8733 and how automatic/manual optimization is a property of the LightningModule. That property also started out as a trainer argument before being migrated to the lightning module. Since this pattern keeps occurring, we should separately understand why it's so appealing to add things to the trainer constructor instead of a more specific component.

Moreover, this one setting controls dataloader behavior for both train & val dataloaders. Do we need more granular control? Do we need two properties, one for training and one for validation? This could make sense as we could have very different epoch counts with features like val_check_interval, where the training epoch count != val epoch count. The property for validation would only apply during trainer.fit as trainer.validate only makes a single pass through the data.

However, the documentation for the test_dataloader: https://github.com/PyTorchLightning/pytorch-lightning/blob/963c26764682fa4cf64c93c5a7572ae0040e9c32/pytorch_lightning/core/hooks.py#L535-L537
Is this a copy/paste issue?

Pitch

• Add a property to the DataHooks class for this in v1.5
• Deprecate the Trainer argument for this in v1.5
• Remove the Trainer argument in v1.7

Benefits:

• Simplify the Trainer constructor (one fewer argument)
• Keep the data loader management in one place instead of two (at the DataHooks level)

Alternatives

Keep as is?

Additional context

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