An instance of SaveConfigCallback should only save the config once

[mauvilsa]

What does this PR do?

When using LightningCLI(.., run=False) it is very common to call multiple Trainer methods, e.g. fit then test. In the second method call, the SaveConfigCallback attempts to save the config again, which by default would fail because it refuses to overwrite. This makes it necessary to have save_config_overwrite=True even though one might not want this. This can be observed in some places, for example #1715 (comment).

This change is so that a single instance of SaveConfigCallback only attempts to save the config once, avoiding the problem described before. I would consider this to be a bug fix.

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

None

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?
• [n/a] 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:

• 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]

I think I agree with the changes. However, I don't understand this sentence:

This makes it necessary to have save_config_overwrite=True even though one might not want this.

In order to reproduce, I think it makes sense to have run=False, overwrite=False, because we want to save the config of the initial fit/test. If we were to overwrite, we would end up with the config of the last ever called fit/test, but that would make it not reproducible.

The logic in the code of this PR however makes sense to me, and I think it implements what I mean.

[tchaton]

LGTM !