fix: raise error when folder-based datasets are loaded without data_dir or data_files

[ArjunJagdale]

Related Issues/PRs

#6152

---

What changes are proposed in this pull request?

This PR adds an early validation step for folder-based datasets (like audiofolder) to prevent silent fallback behavior.

Before this fix:

• When data_dir or data_files were not provided, the loader defaulted to the current working directory.
• This caused unexpected behavior like:
  • Long loading times
  • Scanning unintended local files

Now:

• If both data_dir and data_files are missing, a ValueError is raised early with a helpful message.

---

How is this PR tested?

• Manual test via load_dataset("audiofolder") with missing data_dir
• Existing unit tests (should not break any)
• New tests (if needed, maintainers can guide)

---

Does this PR require documentation update?

• No. You can skip the rest of this section.

---

Release Notes

Is this a user-facing change?

• Yes. Give a description of this change to be included in the release notes for users.

Adds early error handling for folder-based datasets when neither data_dir nor data_files is specified, avoiding unintended resolution to the current directory.

What component(s), interfaces, languages, and integrations does this PR affect?

Components:

• area/datasets
• area/load

---

How should the PR be classified in the release notes? Choose one:

• rn/bug-fix - A user-facing bug fix worth mentioning in the release notes

---

Should this PR be included in the next patch release?

• Yes (this PR will be cherry-picked and included in the next patch release)

[lhoestq]

Great ! Since this logic is specific to one builder class maybe this check can be in the class definition ? I think you can put it in FolderBasedBuilder's _info() method.