[2/4] Add DeviceStatsMonitor callback

[daniellepintz]

What does this PR do?

Adds DeviceStatsMonitor callback which calls the get_device_stats function in Accelerator interface. DeviceStatsMonitor will replace GPUStatsMonitor and XLAStatsMonitor

Part of #9032

Does your PR introduce any breaking changes? If yes, please list them.

Before submitting

• Was this discussed/approved 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 internal minor changes/refactorings)

PR review

Anyone in the community is welcome to review the PR.
Before you start reviewing make sure you have read Review guidelines. In short, see the following bullet-list:

• 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

Did you have fun?

Make sure you had fun coding 🙃

[daniellepintz]

thanks so much @kaushikb11 for fixing the tpu test failure!!!

[daniellepintz]

[RESOLVED] Does anyone have any suggestions to solve the mypy error in a clean way? https://github.com/PyTorchLightning/pytorch-lightning/pull/9712/checks?check_run_id=3875669398

[daniellepintz]

[RESOLVED - we decided pl_module.device makes more sense since we might want stats for non-root device] Also, what do people think about using root_device vs pl_module.device?

[daniellepintz]

Also is there anything I can do about the "Abstract method not overridden" Python errors? https://deepsource.io/gh/PyTorchLightning/pytorch-lightning/run/36c8f43c-2cab-40b2-b81f-6edd6c439a43/python/PYL-W0223

Do we definitely want LightningLoggerBase to be an ABC? cc @carmocca

[carmocca]

Do we definitely want LightningLoggerBase to be an ABC? cc @carmocca

It might be okay to remove - it was added without a strong argument for it: #605

perhaps @justusschock or @awaelchli have a preference?

[awaelchli]

I think we don't want the LoggerBase to be a logger that people can instantiate and use. That would be silly:

trainer = Trainer(logger=LightningLoggerBase())

or what would be the benefit?

[daniellepintz]

Nevermind, I agree that doesn't make sense. The initial motivation was to fix a DeepSource error from a test class, but we can consider this resolved.