Nglui

Changed

- **StateBuilder**: When providing a URL but not a client, you will now get a warning that target-site cannot be inferred.

Fixed

- Small changes to work in python 3.8, because the only issue was type hints.

Fixed

- Removed debug statements left in after testing.

Added

- **StateBuilder** : Added a `skeleton_source` and `skeleton_shader` parameter to SegmentationLayerConfig to put skeleton info into the state.
- **EasyViewer** : Added `set_skeleton_source` and `set_skeleton_shader` methods to EasyViewer.

Fixed

- **SegmentProperties** : Better handling of null values (nan, None, and empty strings) in tag columns, all of which are ignored.
- Various requirements updates that were necessary.

Added

- **SegmentProperties** : Added a `.to_dataframe()` method for segment properties to convert them back to a dataframe.

Added

- **SegmentProperties** : A new module has been created to build Segment Properties, which are a method of organizing information about segment IDs in a Neuroglancer segmentation layer.
Segment Properties only work with the newer Google/Spelunker branches, and offer a way to add browsable and searchable metadata to segments that can be viewed in the viewer.
- **StateBuilder** : Various changes were made to support segment properties from both explicit URLs and from data-driven mapping. See documentation for details.

Changed

- **StateBuilder** : Image layers for spelunker now use native contrast controls.

Fixed

- **StateBuilder** : Improvements to the use of `target_site` and `url_prefix` in general when both creating statebuilder and rendering states. In particular, the values in `render_state` should now correctly take precendence over the values in `StateBuilder` when both are set.

Added

- The **Parser** function `annotation_dataframe` now has a parameter `expand_tags` that will create a boolean column for each tag in each annotation layer indicating if the annotation has that tag or not.

Fixed

- **Parser** : Fixed bugs in getting bounding box annotations and in getting linked segmentations for Spelunker-style states.