[efegokdemir]

Summary

Replace the repeated unused_crate_dependencies Clippy checks with a dedicated cargo machete workflow. This makes dependency checking workspace-wide and includes dependencies used by tests and benches, as requested in #11166.

Changes

• Add a dedicated Dependencies workflow using pinned cargo-machete 0.9.2.
• Run cargo machete --with-metadata for the complete workspace.
• Remove the duplicated dependency-lint invocations from the Arrow and PyArrow Clippy jobs while retaining their normal Clippy checks.

Testing

• cargo machete: passed; no unused dependencies detected.
• yamllint -d relaxed .github/workflows/arrow.yml .github/workflows/integration.yml .github/workflows/dependencies.yml: passed with existing line-length warnings.
• git diff --check: passed.
• cargo machete --with-metadata: attempted locally, but the offline environment lacks cached workspace dependencies; the command completed with dependency-download errors. The CI workflow runs in a populated container and uses the same command.

Notes

• The workflow uses the same pinned taiki-e/install-action already used by Apache DataFusion for this tool.
• AI assistance was used during preparation; the changes and validation were reviewed by the submitting contributor.

Fixes #11166

[Jefffrey]

closing as i suspect its part of AI spam:

not to mention this was not even verified locally properly:

cargo machete --with-metadata: attempted locally, but the offline environment lacks cached workspace dependencies; the command completed with dependency-download errors. The CI workflow runs in a populated container and uses the same command.

i explicitly mention in my original issue it throws false positives, so it seems there was no actual attempt made at verifying this PR locally