Updates docs to clarify train_dataloader usage by batch_size_finder (#13171)

* batchfinder clarification

* fix according to precommit hook

CHANGELOG.md

@@ -119,7 +119,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Deprecated `pytorch_lightning.core.lightning.LightningModule` in favor of `pytorch_lightning.core.module.LightningModule` ([#12740](https://github.com/PyTorchLightning/pytorch-lightning/pull/12740))
 
 
-- Deprecated `Trainer.reset_train_val_dataloaderrs()` in favor of `Trainer.reset_{train,val}_dataloader` ([#12184](https://github.com/PyTorchLightning/pytorch-lightning/pull/12184))
+- Deprecated `Trainer.reset_train_val_dataloaders()` in favor of `Trainer.reset_{train,val}_dataloader` ([#12184](https://github.com/PyTorchLightning/pytorch-lightning/pull/12184))
 
 ### Removed
 
@@ -241,7 +241,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Fixed issue where the CLI could not pass a `Profiler` to the `Trainer` ([#13084](https://github.com/PyTorchLightning/pytorch-lightning/pull/13084))
 
 
-- Fixed logging on step level for eval mode ([#12184](https://github.com/PyTorchLightning/pytorch-lightning/pull/12184))
+- Fixed logging's step values when multiple dataloaders are used during evaluation ([#12184](https://github.com/PyTorchLightning/pytorch-lightning/pull/12184))
 
 
 -

docs/source/advanced/training_tricks.rst

@@ -143,6 +143,12 @@ search for batch sizes larger than the size of the training dataset.
         datamodule = LitDataModule(batch_size=32)
         trainer.tune(model, datamodule=datamodule)
 
+    Note that the ``train_dataloader`` can be either part of
+    the ``LightningModule`` or ``LightningDataModule``
+    as shown above. If both the ``LightningModule``
+    and the ``LightningDataModule`` contain a ``train_dataloader``,
+    the ``LightningDataModule`` takes precedence.
+
 .. warning::
 
     Due to the constraints listed above, this features does *NOT* work when passing dataloaders directly