Add typed multiple column panel model

[palewire]

Summary

• Adds a typed MultipleColumnPanel model for MultipleColumnChart.panels.
• Preserves the column name as a Python attribute while serializing to Datawrapper's keyed metadata.visualize.panels object.
• Accepts both the existing list-of-dicts input style and the keyed dict API shape for compatibility.
• Exports MultipleColumnPanel from datawrapper and datawrapper.charts, and updates docs/examples.

Fixes #495.

Evidence and compatibility notes

• Existing multiple-column fixtures already include real metadata.visualize.panels data in the Datawrapper API keyed-dict shape.
• Before this change, deserialized panels were plain dictionaries, so users could not work with a typed panel model.
• The previous serializer also included column inside each panel value; this change uses the column as the API key and serializes only panel configuration fields as the value.
• Unknown panel fields are preserved with extra="allow" so future Datawrapper panel options can round-trip without a wrapper release.
• Legacy list-of-dicts panel input remains supported, including dictionaries containing column.

Validation

• uv run ruff check .
• uv run ruff format --check datawrapper/charts/multiple_column.py datawrapper/__init__.py datawrapper/charts/__init__.py tests/integration/test_multiple_column_chart.py
• uv run pytest tests/integration/test_multiple_column_chart.py -q — 27 passed
• uv run pytest -q — 1108 passed, 3 skipped
• git diff --check

Note: uv run ruff format --check . reports one pre-existing formatting issue in tests/integration/test_scatter_chart.py on origin/main, so I limited the formatter check to the files changed by this PR to avoid unrelated churn.

[Copilot]

Pull request overview

This PR introduces a typed MultipleColumnPanel model to represent MultipleColumnChart.panels, while preserving Datawrapper’s keyed metadata.visualize.panels API shape during serialization and supporting both legacy and API-shaped inputs for backwards compatibility.

Changes:

• Added MultipleColumnPanel Pydantic model and updated MultipleColumnChart to use it for panels, including keyed-dict serialization/deserialization behavior.
• Extended integration tests to cover typed panels, API keyed-dict initialization, and compatibility with the wrapper’s previous keyed-dict output shape.
• Exported MultipleColumnPanel from datawrapper/datawrapper.charts and updated documentation/examples to use the typed model.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
datawrapper/charts/multiple_column.py	Adds MultipleColumnPanel, updates panels typing, conversion, and keyed-dict serialization/deserialization.
tests/integration/test_multiple_column_chart.py	Adds coverage for typed panel behavior and compatibility input shapes.
datawrapper/charts/__init__.py	Exports MultipleColumnPanel from datawrapper.charts.
datawrapper/__init__.py	Exports MultipleColumnPanel from the top-level package.
docs/user-guide/charts/multiple-column-charts.md	Updates examples to use MultipleColumnPanel.
docs/user-guide/api/models.rst	Documents the new MultipleColumnPanel model in the API reference.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[palewire]

Follow-up pushed in f40238c.

• Addressed the convert_panels review feedback: keyed-dict keys are authoritative, including when value-level column conflicts, and unsupported panel item types now raise clear TypeErrors.
• Added focused regression coverage for conflicting keyed/value column, keyed MultipleColumnPanel input, and invalid keyed/list panel items.
• Repaired the failing Python lint/format check by applying Ruff formatting to tests/integration/test_scatter_chart.py, matching the GitHub Actions failure.

Local validation run:

uv sync --frozen --extra dev --extra test --extra mypy
uv run ruff check ./datawrapper ./tests
uv run ruff format --check ./datawrapper ./tests
uv run mypy ./datawrapper --ignore-missing-imports
uv run pytest

Result: Ruff and mypy passed; full pytest passed with 1112 passed, 3 skipped.