Eomaps

🔨 Fixes
- fix minor dependency issues reported on conda-forge

A new minor release that brings a few features and a much improved [documentation](https://eomaps.readthedocs.io)!

🌳 New

- There is a new method [`ScaleBar.set_rotation`](https://eomaps.readthedocs.io/en/dev/generated/eomaps.scalebar.ScaleBar.set_rotation.html) to set the rotation-angle of a scalebar.
- [`m.set_frame(gdf=...)`](https://eomaps.readthedocs.io/en/dev/generated/eomaps.eomaps.Maps.set_frame.html) can now be used to set arbitrary map-boundaries based on `geopandas.GeoDataFrames`
(multipolygon support requires upcoming cartopy v0.24 release)
- There is a new contextmanager [`Maps.delay_draw`](https://eomaps.readthedocs.io/en/dev/generated/eomaps.eomaps.Maps.delay_draw.html) that can be used to delay intermediate draw-events and speed up the creation of complex figures.

📖 Documentation updates and new theme
> 🚀 Big thanks to Fanchengyan for his efforts to improve the documentation!

- Docs now use the [`PyData Sphinx Theme`
](https://pydata-sphinx-theme.readthedocs.io)
- New improved example gallery based on [`myst_sphinx_gallery`](
https://myst-sphinx-gallery.readthedocs.io) ready to be filled with a lot of examples!

🔨 Fixes
- Fix issues with "inherit_position=False" for colorbars
- Make sure cached layers are re-drawn if a colorbar changes the axes-size
- Fix treatment of assigned "search_radius" for `pick` callbacks if 1D brute-force search is used
- Update GitHub actions and documentation dependencies

A first series of bugfixes for [v8.2](https://github.com/raphaelquast/EOmaps/releases/tag/v8.2)

🔨 Fixes
- ❗Fix identification of default shape based on dataset size
- ❗Fix issues with unmanaged artists on inset-maps
- Explicitly check if norm is provided as string before setting data limits
- Provide more descriptive (info) error message if data cannot be plotted
- Avoid using classification bins as labels if a large number of bins is used (>30)
- Make sure to resample colormaps in case more bins than colors are used
- Avoid shape radius estimation if no data is set
- Avoid unnecessary draw-triggers for datasets if the "all" layer is visible
- Remove warning for widgets if backends other than "ipympl" are used

A new minor release with some fixes, many improvements to the internals and a few new features!

🌳 New
- You can now use `m.add_colorbar(layer=..)` to override the layer at which the colorbar is plotted
- There is a new method `m.set_shade_dpi(...)` to override the dpi used by shade-shapes.
(Useful to fix data aggregation resolution for high-dpi exports or to adjust resolution to better suit data density)

🌦️ Changes
- Grid-snapping of the `LayoutEditor` now snaps to **all** edges of the axes
(Useful in case the axis-size is not a multiple of the grid-spacing)
- If multiple axes are moved, the first selected object acts as anchor
- ❗ The default matplotlib keymap `p : "toggle pan/zoom tool"` is now active again

🌩️ Internals (no effects on the public API)
- Unit-tests now use image-comparison to ensure stable image export results
- ❗ The `Maps` class has been refactored (`Maps` and `MapsBase`) to make the code more accessible and easier to handle.

🔨 Fixes
- fix handling of axes on combined layers in the LayoutEditor
- allow custom scatter_points marker shapes for "mark" callback
- fix issues with identification of grid-label position with projections that create non-intersecting grid-lines
- always print basic exception info on issue during layer-change action
- make sure to trigger a figure-update after adding a scalebar
- address pyproj warnings for pyproj>=3.5

🔨 Fixes
- ❗ Fix colorbars on different layers only show after additional draw
- Handle numpy linalg errors resulting from singular colorbar axes

A minor release that brings a new plot-shape!

> Make sure to also checkout the release-notes for [EOmaps v8.0](https://github.com/raphaelquast/EOmaps/releases/tag/v8.0.2)!

🌳 New

New hexbin shape
There is a new shape to visualize your data as [hexagonal-bin plots](https://eomaps.readthedocs.io/en/dev/api_data_visualization.html#shp-hexbin)!
python
m.set_shape.hexbin(size=(20, 40), aggregator="mean")

<img src="https://github.com/raphaelquast/EOmaps/assets/22773387/2e8429f1-6aa4-4b5e-a8a9-a3370ab76549" width=70%>

🔨 Fixes
- ❗ fix using multiple sizes with `m.set_shape.scatter_points(size=...)`
- Fix scroll events should be captured in Jupyter Notebooks if the LayoutEditor is active
- Fix placement of `InsetMaps` indicator-lines if multiple boundary intersections are found
- Fix clipping issues if `m.get_extent()` is used with a crs other than the plot-crs
- Fix using `m.add_gridlines` with uneven number of explicit lon/lat values
- Unify treatment of `ellipses` and `geod_circle` shapes
- Fix using kwargs to style the boundary of the `peek_layer` callback area
- Fix using "hist_size=None" when adding colorbars
- Fix hiding histogram axes with hist_size=0
- Fix identification of layers that contain only dynamic artists
- Fix issues with scalebars outside the axes clip boundary
- Allow passing an explicit norm to `m.plot_map` if no classification is used
- Allow using `bbox=None` with annotations to to hide the bounding-box
- Identification of the directions for `m.add_gridlines(where="tblr")` has been improved