Fix integration of LearningRateFinder with EarlyStopping

[SkafteNicki]

What does this PR do?

Fixes #19575
Fixes #20355

When using LearningRateFinder there is currently a bug that makes is so that on_train_epoch_end hooks are executed as one of the first things when starting fitting afterwards. The reason for this seems to be that the data_fetcher returns StopIteration after the learning rate finding, which makes the self.advance() method in fit loop return immediately

pytorch-lightning/src/lightning/pytorch/loops/fit_loop.py

Lines 207 to 223 in f067626

	def run(self) -> None:
	self.setup_data()
	if self.skip:
	return
	self.reset()
	self.on_run_start()
	while not self.done:
	try:
	self.on_advance_start()
	self.advance()
	self.on_advance_end()
	except StopIteration:
	break
	finally:
	self.on_iteration_done()
	self._restarting = False
	self.on_run_end()

afterwards the self.on_advance_end is executed which then calls the on_train_epoch_end hook from EarlyStopping which then fails if we are monitoring a validation metric, because the validation loop have not been called yet (which is expected).

Solution is to reset data_fetcher to None such that it is re-initialized during the actual fit loop.

Before submitting

• Was this discussed/agreed 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:

Reviewer checklist

• 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

---

📚 Documentation preview 📚: https://pytorch-lightning--21056.org.readthedocs.build/en/21056/