fix gpus default for Trainer.add_argparse_args

[awaelchli]

What does this PR do?

Fixes #6549
Fixes #6263

The default for gpus and tpu_cores was a function, but it should be None as in the Trainer.
This fixes a pickle issue, when the function is passed into the model via Model(**vars(args)).
It would prevent unpickling outside pytorch lightning (or when PL code changes).

Newly added test case fails on master.

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 update the CHANGELOG? (not for typos, docs, test updates, or internal minor changes/refactorings)

PR review

Anyone in the community is free to review the PR once the tests have passed.
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 🙃

[awaelchli]

this is the fix.
we should not attach a function here, because if we pass it into our hparams, it will land in the checkpoint and risk not being able to unpickle.

[ethanwharris]

LGTM 😃

[carmocca]

Do you have any idea why the default was a function?

[awaelchli]

@carmocca totally no clue. Maybe someone confused it with the type, which can be a function that argparse uses to covert the input.