Eomaps

Another round of bug-fixes for [v7.3](https://github.com/raphaelquast/EOmaps/releases/tag/v7.3).

🌳 New
- `MapsGrid` objects now have dedicated `.show()` and `.snapshot()` methods

🔨 Fixes
- ❗ fix restoring visible layer in interactive backends after using `m.savefig()`
- ❗ fix issues with InsetMaps in `ipympl` backend
- fix re-drawing spines with `m.set_frame()` if corners are not rounded
- fix colorbar should use figure directly instead of relying on `plt.gcf`
- fix re-fetching backgrounds if colorbars change axes sizes
- fix parsing "maxsize" argument for raster-shape in companion-widget
- fix treatment of weighted colorbar histograms with (out_of_range_vals="masked")

A second bugfix release for [v7.3](https://github.com/raphaelquast/EOmaps/releases/tag/v7.3) that brings some important fixes for Jupyter Notebooks, custom WebMap services and RGB/RGBA composite plots.

🌳 New

- Style kwargs are now supported when adding multiple preset-features in one go, for example:
python
m = Maps()
m.add_feature.preset("coastline", "ocean", "land", alpha=0.5)

- You can now specify additional wms properties when using custom WebMap services, for example:
python
my_wms = m.add_wms.get_service("service-url", service_type="wms", version=1.3.0)


🌦️ Changes
- The figure-header added by the **Jupyter-notebook** `ipympl` backend is now hidden by default.
To revert back to the old behavior (and enable the header), use:
python
%matplotlib widget
m = Maps()
m.f.canvas.header_visible = True


🔨 fixes
- ❗ fix issues with **RGB/RGBA composite plots**
- ❗ fix **UserDefined Classification** issue: max-bin replaced by data-max
- ❗ fix using transparent snapshots in IPython/Jupyter Notebooks
- fix duplicated figures shown in **jupyter notebooks** inline backend
- fix re-fetching backgrounds and spines with `m.set_frame()`
- fix using `m.set_frame()` in jupyter notebook inline backend
- improve syntax for disabling draw or update in the BlitManager
- improve handling of issues with "tight_bbox" redraws
- avoid logging during draw on higher levels
- avoid turning off interactive mode with the jupyter notebook inline backend
- allow passing kwargs to custom WebMap services

A first series of bugfixes for [v7.3](https://github.com/raphaelquast/EOmaps/releases/tag/v7.3).

🌦️ changes
- InsetMap indicator lines are now only drawn between the boundaries of the InsetMap and its extent.

You can add markers at the connection points with:

python
m_inset.add_indicator_line(m, marker="o", markersize=10)

<img src=https://github.com/raphaelquast/EOmaps/assets/22773387/57c07798-7a1c-4e38-bb55-2af72de37ef3 width=50%/>


🔨 Fixes
- fix re-drawing backgrounds on add/remove gridlines and gridlabels
- fix re-drawing backgrounds on `m.set_frame`
- fix triggering on_layer callbacks defined on InsetMaps
- fix custom text positioning with `m.add_title`
- fix handling of multiple inset-indicator lines
- fix passing InsetMap boundary colors as RGBA tuples

A new minor release that brings a lot of fixes and some nice new functionalities!

🌳 New
- There is a new convenience method to quickly update the properties of the frame of a map:
- set boundary line properties and **use rounded corners for map boundaries**
- set background patch color

python
m = Maps()
m.set_frame(rounded=0.5, edgecolor="r", linewidth=2, facecolor="0.5")



- `m.add_annotation()` and `m.add_marker()` now support the `"update"` kwarg to override update triggers
(useful to avoid unnecessary updates if multiple dynamic annotations/markers are added in a callback)
- `m.add_gdf()` now supports the "permanent" kwarg to control if artists are updated on each event
or added as background-artists
- The [🚀 Contribution guide](https://eomaps.readthedocs.io/en/latest/contribute.html) contains a new chapter on how to build the docs!

🌦️ Changes
- ❗ Using `m.show()` no longer activates the layer associated with the calling Maps object.
(use `m.show_layer(m.layer)` to show the layer of a selected Maps object!)
- Activating the `AnnotationEditor` now clears all currently visible temporary artists and disables all callbacks


🔨 Fixes
- ❗fix zooming datasets with individual shape sizes for each datapoints
- ❗fix using the save-button in the normal matplotlib toolbar to save a figure
- ❗fix custom axis background patch color is only visible on figure export
- ❗fix grid-labels are not properly updated on pan/zoom events
- fix identification of layers that only exist on `InsetMaps`
- fix annotation editor clears annotation text on companion-widget init
- fix all-nan slice warnings on vmin/vmax estimation if no data-values are provided
- fix single geod_circle radius parsed to 1D array in setter
- fix using `m.add_gdf(permanent=False)` to add dynamically updated geometries
- fix annotation editor issues if annotation is not attached to the figure
- fix `m.get_extent()` should not return coordinates outside the crs bounds
- fix closing companion-widgets of child-maps on figure close
- fix clearing snapshot cells with `m.show()`
- fix `m.util.layer_selector()` should remain responsive during toolbar actions
- fix using `m.cb.<method>.set_execute_during_toolbar_action()` on child Maps objects
- improve handling of errors during snapshot creation
- remove `rtree` and add default python version to development env file
- fix restoring axes backgrounds in gui after savefig

A first bugfix release for v7.2 that also brings some nice performance improvements!

> Make sure to also check the [release-notes for EOmaps v7.2](https://github.com/raphaelquast/EOmaps/releases/tag/v7.2)!

🔨 Fixes
- ❗ Fix issues with aggregated datasets outside the currently visible extent
- ❗ Fix displacement of grid-labels on figure export with dpi other than the native dpi
- ❗ Fix dataset export issues with `"agg"` backend on some of the plot-shapes
- general performance updates for background-fetching and caching
- avoid activating the companion widget when webagg backend is active

🌳 New
- New method for raster-data aggregation: `"mode"` (e.g. most common value)
- The behavior of callbacks during toolbar actions can now be set via
`m.cb.<method>.set_execute_during_toolbar_action()`
- New `.remove()` method for grid-lalbels

🌦️ Changes
- To avoid issues with `"webagg"` derived backends, custom `"move"` callbacks no longer trigger automatic updates of the plot.
- ❗ Explicitly call `m.BM.update()` in custom move-callbacks to trigger an update!

A mew minor release that brings a lot of improvements for raster-data visualization!

🌳 New
Major improvements for raster-data visualization!
The raster-shape (e.g. `m.set_shape.raster`) now has data-aggregation capabilities!

> This means that the `raster` shape can now be used also for **arbitrarily sized datasets** (as long as they fit in memory)
> (so far tested with datasets up to 2 billion datapoints)

In short, this means:
- Datasets larger than `maxsize` will be aggregated prior to plotting to avoid slow load-times for extremely large datasets
- Aggregation can be turned off using `maxsize=None` (e.g. to get the behavior prior to v7.2)
- By default, only datasets larger than 5 million datapoints will be aggregated prior to plotting

For more details, check the [docs for the raster-shape](https://eomaps.readthedocs.io/en/latest/generated/eomaps.shapes.Shapes.raster.html#eomaps.shapes.Shapes.raster)!

🔨 Fixes
- ❗ avoid activating matplotlibs interactive mode by default
- fix identification of data-limits and colormap for GeoTiffs

🌦️ Changes
- `raster` is now used as default plot shape also for very large datasets