Move tests/samples to HF?

[asumagic]

What does this PR do?

WIP. Proof-of-concept of using a submodule towards HF using LFS, which has a few practical advantages:

• HF limits are much more generous than GitHub's for LFS
• Better integration with git
• Default clone commands will not clone this submodule
• Clone commands with the recursive submodules option, by default, will only fetch the latest version of the file thanks to LFS
• We retain versioning for the samples files. Realistically, we could be fine without, but it's still a bonus if it doesn't pollute the main repo history in a significant way and keeps defaults sane

The obvious downgrade is that it makes recipe PRs more annoying.

Before submitting

• Did you read the contributor guideline?
• 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?
• Does your code adhere to project-specific code style and conventions?

PR review

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
• Confirm that the changes adhere to compatibility requirements (e.g., Python version, platform)
• Review the self-review checklist to ensure the code is ready for review

[asumagic]

The PR seems to work in my testing. Needs more discussion to see if people are OK with this.

Then, we will need migrating that new repo under the speechbrain/ org, documenting the procedures, updating the license details, etc.

[pplantinga]

The obvious downgrade is that it makes recipe PRs more annoying.

This is only for recipe PRs that add data to the test samples right? And the change to the process is simply to open a concurrent PR to the HF test samples repo right?

[asumagic]

The obvious downgrade is that it makes recipe PRs more annoying.

This is only for recipe PRs that add data to the test samples right? And the change to the process is simply to open a concurrent PR to the HF test samples repo right?

Correct, but one problem that just occurred to me is that we need to think about the flow for updating the submodule from a PR.

Apparently, submodules allow tracking branches nowadays which might make sense here.