Lint overhaul (pylint to ruff)

[akx]

This PR looks a bit big, but it's pretty straightforward if you look at the commits in isolation.

Basically, it replaces pylint and isort with the ruff linter, and re-enables linting in CI (now that it's not screaming an awful lot – I hope @erogol has found some peace of mind (ref. 8c1d8df)). The pylint configuration was converted with ttps://github.com/akx/pylint-to-ruff.

• pre-commit is also updated, but currently running pre-commit run --all-files yields a ton of changes.
• Some ruff rules are not enabled at this point at all because they're kind of noisy, but they could be enabled in future PRs one by one.
• This PR supersedes Update pre-commit tools #2992.
• I found [Bug] compute_attention_masks.py doesn't work #3002 while working on this, so hey, there's clearly a point to this.

[erogol]

Hey @akx. Can you explain why ruff over pylint a little?

[akx]

@erogol

• Ruff is a lot (up to 100x) faster than pylint (so linting isn't frustrating and can be done in e.g. pre-commit)
• It supports most of what Pylint does, and more by way of non-Pylint rules: Implement Pylint astral-sh/ruff#970
• Even pylint uses ruff: https://github.com/pylint-dev/pylint/blob/main/pyproject.toml#L141

[akx]

@erogol Rebased on current dev.

[akx]

@erogol Rebased once again... 🙂

[akx]

... and again, following #3149...

[akx]

... and again. This also includes #3170 because it'd otherwise cause a lint failure.

[akx]

@erogol Could this get merged now? You can always disable some lints if they end up being too noisy...

[akx]

Rebased for the new month. The finetuning example that was merged in #3296 was pretty messy...

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions. You might also look our discussion channels.

[eginhard]

I've merged this and the check_skip PR into our fork, thank you! :)