Jupyter-scatter

- Feat: automatically assign gray to undefined/NA values for categorical data ([186](https://github.com/flekschas/jupyter-scatter/issues/186))
- Fix: properly undefined/NA values in categorical data ([185](https://github.com/flekschas/jupyter-scatter/issues/185))
- Fix: tooltip issue with non-existing tooltip ([187](https://github.com/flekschas/jupyter-scatter/issues/187))
- Fix: ensure `tooltip_histogram == False` works as expected
- Fix: ensure a static opacity is not shown in the tooltip

- Feat: add support for brush and rectangle lasso types.
- Feat: expose lasso selection polygon via `scatter.widget.lasso_polygon_selection`
- Feat: enable merging of lasso selection upon holding down the [meta key](https://developer.mozilla.org/en-US/docs/Web/API/KeyboardEvent/metaKey) and deselected points upon holding down the [alt key](https://developer.mozilla.org/en-US/docs/Web/API/KeyboardEvent/altKey).
- Feat: allow specifying which properties should feature a histogram via `scatter.tooltip(histograms=['color', 'property_a'])`
- Fix: background histogram render bug
- Fix: ensure NAs are skipped for the histogram
- Fix: ensure that the tooltip is hidden on lasso start
- Fix: ensure that the color, opacity, and size order is reset when one switches from a categorical to a continuous color map. The reset is skipped when `order == 'reverse'`.

- Feat: add support for showing a scatter plot in full-screen mode. In full-screen mode you can easily customize and up-scale the resolution for better exporting in PNG.
- Feat: add ability to change the point size zoom scale function via `scatter.size(scale_function='asinh')`
- Feat: add `scatter.camera(is_fixed=True)` to allow fixing the camera (i.e., disable pan/zoom) [176](https://github.com/flekschas/jupyter-scatter/issues/176)
- Feat: add CLI for quick-starting a demo via `uvx jupyter-scatter demo`
- Fix: properly alpha blend when saving a view as PNG
- Chore: switch from ESLint+Prettier to Biome [170](https://github.com/flekschas/jupyter-scatter/pull/170)
- Chore: migrate from hatch to uv [169](https://github.com/flekschas/jupyter-scatter/pull/169)
- Chore: bump minimum Python version to 3.9

- Fix: revert back to `pd.api.types.is_string_dtype` from `pd.StringDType` because the two function, albeit looking semantically identical, do not do the same thing. The former detects object and string types while the latter only strictly detects string types. This is confusing as `pd.Series(['a', 'b'])` is of type `object`. So `pd.StringDtype.is_dtype(pd.Series(['a', 'b'])) == False` but `pd.api.types.is_string_dtype(pd.Series(['a', 'b'])) == True`. [Wat](https://www.destroyallsoftware.com/talks/wat)?!

- Feat: add contour line annotations [163](https://github.com/flekschas/jupyter-scatter/issues/163)
- Refactor: rename the `reset_view` argument of `scatter.data()` to `zoom_view` for clarity [162](https://github.com/flekschas/jupyter-scatter/issues/162)
- Docs: add API docs for `reset_scales`, `zoom_view`, and `animate` of `scatter.data()`
- Fix: don't use external view domain (only the view) [161](https://github.com/flekschas/jupyter-scatter/issues/161)

- Fix: re-enable point transition and make it explicit via `scatter.options(transition_points=True, transition_points_duration=3000)`
- Fix: Correctly render axes grid lines upon resizing
- Fix: Reapply point color map when setting `color_by`
- Fix: Do not set tooltip titles if the corresponding elements are undefined.