Added some more potentially robust ways to do learning rate tuning

[varchasgopalaswamy]

What does this PR do?

As discussed in #1767, the learning rate tuner suggests the minimum gradient of the loss via torch.argmin(torch.gradient(losses)[0]). This makes it a little sensitive to a noisy loss vs LR landscape. I added a few methods discussed here, and found they worked pretty well for my use cases (see example below, where gradient-based estimation picks out a very low LR).

Defaults are set so no breaking changes are introduced.

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--19867.org.readthedocs.build/en/19867/