Paintera-0.8.2 fixes issues related to opening and (de-)serializing datasets.
--------
Bug Fixes
- Load label sources without `maxId` attribute, choose one of the following when opening or de-serializing dataset (165):
1. Do not use id service: Cannot paint with new id or merge/detach fragments.
2. Specify `maxId` manually: User's responsibility to be conistent with data.
3. Scan data for `maxId`: might still be inconsistent with label-block-lookup.
- Ensure Java process stops when closing Paintera (160, 166):
- Make dataset discovery handler/main thread daemon thread.
- Only scan for datasets if selected directory is a valid n5 container (has `attributes.json` with `"n5"` attribute).
- Stop dataset discovery when `Cancel` is pressed in dialog.
- Make all `ExecutorService`s in `PainteraBaseView` daemon.
- Use writers only for HDF5 sources (168, 166)
- Add option to generate 3D meshes from entire dataset if no label-block-lookup is provided
- (De-)Serialize `LabelSourceState` agnostic of `N5` -- keep fallback but emit warning if appropriate (167, 166).
- Remove fat profile (use [`jgo`](https://github.com/scijava/jgo) instead)