Deleting a dataset doesn't remove it from the Hub

## Describe the Bug / Missing Feature

`Delete Dataset` in LeLab only removes the local copy under `~/.cache/huggingface/lerobot/`.
If the dataset was already pushed to the Hub, it's still sitting there afterwards.
LeLab gives no way to remove it, you have to go delete the repo manually on huggingface.co.

## Steps to reproduce

1. Record a dataset and push it to the Hub.
2. Go delete it from LeLab (dataset list -> delete).
3. Check the Hub — the repo is still there.

## Expected behavior

Deleting a dataset that's been pushed to the Hub should give the option to
also delete the Hub repo, not just wipe the local cache.

## Proposed fix

`handle_delete_dataset` (`lelab/record.py`) would need to call `huggingface_hub.delete_repo(repo_id, repo_type="dataset")` when the dataset is Hub-hosted, probably behind its own confirmation step in the UI since it's not reversible (e.g. type the repo name to confirm, similar to GitHub's repo delete flow).

Start in lelab/record.py at handle_delete_dataset and trace how the dataset-list delete action identifies Hub-hosted datasets. Review the huggingface_hub.delete_repo call described in the issue and the existing delete confirmation flow. Done means deletion offers a separate confirmation for the Hub repo and removes it when confirmed, while local-only deletion still works.

Written by the indexing model from the issue text.