Nglui

Changed

- Switched from numpy `isnan` to pandas `isnull` in StateBuilder. This allows nullable pandas Int64 dtypes to work as columns

Added

- Grouped annotations in nglite base and EasyViewer's annotation module.
This feature lets you create annotation groups in the Seung-lab branches of Neuroglancer, where multiple annotations are related to one another.
It works by passing a list of already-created annotations and getting a new "CollectionAnnotation" that groups them together.

- Grouped annotations in StateBuilder. PointMapper, SphereMapper, and LineMapper objects can take a 'group_column' value.
Data in this column is intended to be numeric or NaN, and rows that share the same value are grouped together.
At the moment, this feature only works within individual mapper objects.

- Two options for linked segmentation ids on grouped annotations:
`gather_linked_segmentations` assigns all linked ids of objects within the group to the group annotation itself (True by default).
`share_linked_segmentations` will add all linked objects within the group to all annotations within the group (False by default).

- `array_data` option on AnnotationLayerConfig for simple cases where you just want to map Nx3 arrays to points, a pair of Nx3 arrays to lines, or an Nx3 + N array to spheres (centers+radii).

Fixed

- Bug in GL Shader that caused StateBuilder to fail when `constrast_controls` was set to True.

Added

- Custom GL Shaders (see [documentation](https://github.com/google/neuroglancer/blob/a12552d03844fb6092cf300171d2f2077b3960e2/src/neuroglancer/sliceview/image_layer_rendering.md)) for image and segmentation layers can be set in EasyViewer.

- JSON State Server can be set in EasyViewer.

- Added `contrast_controls` argument to `statebuilder.ImageLayerConfig` to provide brightness/contrast controls to layer through GL Shader.

- Added `state_server` argument to `statebuilder.StateBuilder` to pre-set state server endpoints.

Fixed

- Removed unnecessary parts of the included neuroglancer module (`nglite`) for cleanliness.

This update significantly changed the underlying nature and goals of the Neuroglancer Annotation UI project.

Added

- StateBuilder module for rule-based method of generating Neuroglancer states from Pandas DataFrames.

- Added various features to make use of specific aspects of the Seung-lab Neuroglancer fork.

Changed

- Removed dependency on neuroglancer python package, which was not designed for the Seung-Lab Neuroglancer fork and had some difficult install requirements for features that were unnecessary for this project demanded.

Removed

- Dynamic state management via python server.