dev: initial implementation of notebook-snapshot extension

Manual-commit notebook snapshot extension (JupyterLab 4):

Backend
- Storage adapter layer (Local + S3 via boto3, pip extra)
  - Dumb put/get/list/delete contract; key validation
  - S3Storage: lazy boto3 import, S3ConnectionError with friendly
    messages + connectivity check at startup
- SnapshotStore
  - Strip code outputs/execution_count, MD5 on cleaned JSON
  - Gzipped envelope (id/timestamp/name/description/hash/size/notebook)
  - Append-only manifest.json (rebuildable from version files)
  - SnapshotUnchangedError when new hash matches most recent version
- REST API (commit / list / content) under /snapshot/ namespace
- traitlets config (storage_type, local_root, s3_* + S3_* env fallback)

Frontend
- snapshot:commit command (toolbar button + command palette)
- Dialog for name + description
- SnapshotPanel (left sidebar timeline) + DiffWidget (main area)
- Cell diff: id-match first, LCS fallback; jsdiff for line diff
- Restore via model.fromJSON() + context.save() (no refresh)
- All AGENTS.md hard conventions enforced

Tests: 33 backend pytest passing (storage + store + routes)

Docs: AGENTS.md, design.md, README.md synced with implementation.

Co-Authored-By: Claude <noreply@anthropic.com>

RELEASE.md

@@ -0,0 +1,88 @@
+# Making a new release of snapshot
+
+The extension can be published to `PyPI` and `npm` manually or using the [Jupyter Releaser](https://github.com/jupyter-server/jupyter_releaser).
+
+## Manual release
+
+### Python package
+
+This extension can be distributed as Python packages. All of the Python
+packaging instructions are in the `pyproject.toml` file to wrap your extension in a
+Python package. Before generating a package, you first need to install some tools:
+
+```bash
+pip install build twine hatch
+```
+
+Bump the version using `hatch`. By default this will create a tag.
+See the docs on [hatch-nodejs-version](https://github.com/agoose77/hatch-nodejs-version#semver) for details.
+
+```bash
+hatch version <new-version>
+```
+
+Make sure to clean up all the development files before building the package:
+
+```bash
+jlpm clean:all
+```
+
+You could also clean up the local git repository:
+
+```bash
+git clean -dfX
+```
+
+To create a Python source package (`.tar.gz`) and the binary package (`.whl`) in the `dist/` directory, do:
+
+```bash
+python -m build
+```
+
+> `python setup.py sdist bdist_wheel` is deprecated and will not work for this package.
+
+Then to upload the package to PyPI, do:
+
+```bash
+twine upload dist/*
+```
+
+### NPM package
+
+To publish the frontend part of the extension as a NPM package, do:
+
+```bash
+npm login
+npm publish --access public
+```
+
+## Automated releases with the Jupyter Releaser
+
+The extension repository should already be compatible with the Jupyter Releaser. But
+the GitHub repository and the package managers need to be properly set up. Please
+follow the instructions of the Jupyter Releaser [checklist](https://jupyter-releaser.readthedocs.io/en/latest/how_to_guides/convert_repo_from_repo.html).
+
+For the release workflows in this repository, make sure GitHub is configured with:
+
+- a `release` environment
+- an `APP_PRIVATE_KEY` secret
+- an `APP_ID` repository variable
+
+When using [npm trusted publishing](https://docs.npmjs.com/trusted-publishers), `NPM_TOKEN` is not required (and trusted publishing is recommended). Configure `NPM_TOKEN` only if you are publishing without trusted publishers.
+
+Here is a summary of the steps to cut a new release:
+
+- Go to the Actions panel
+- Run the "Step 1: Prep Release" workflow
+- Check the draft changelog
+- Run the "Step 2: Publish Release" workflow
+
+> [!NOTE]
+> Check out the [workflow documentation](https://jupyter-releaser.readthedocs.io/en/latest/get_started/making_release_from_repo.html)
+> for more information.
+
+## Publishing to `conda-forge`
+
+If the package is not on conda forge yet, check the documentation to learn how to add it: https://conda-forge.org/docs/maintainer/adding_pkgs.html
+
+Otherwise a bot should pick up the new version publish to PyPI, and open a new PR on the feedstock repository automatically.