ImageJ2 and BigDataViewer and ...

After a discussion with @ctrueden in person, we found that it might make sense to open an issue to discuss what steps would be required on ImageJ2 side to allow the following relatively simple use-case using ImageJ2 + BigDataViewer:

## Region of Interests in BigDataViewer (2D) with very large images

### Use Cases
#### Display of Segmentation Results
Allow display `ImgLabeling`s alone or on top of other images.

#### Manual Annotation
Allow to manually draw region of interests (`Polygon`, `Rectangle`, `Freehand`,...) in `BigDataViewer`.

#### Measurements
For each annotation we want to measure features of the roi and output the result in a `Table`

#### Processing
We want to be able to process only the selected regions of interests. If the output is some `RAI` again only the updated regions should be updated in the BDV.

### What's missing?
1. BigDataViewer in 2D mode
2. Allow manual creation of ROIs in BigDataViewer 2D mode.
3. Allow ImageJ2-Commands to consume these ROI information.
4. Allow ImageJ2-Commands to update the entire image or only the processed pixels in the
displayed image.

All of this is somehow related to `RichImg`, `imglib2-roi` and `SCIFIOImgPlus`. However, a lot of this functionality is already in place (`Cache`, `CellImg`, https://github.com/bigdataviewer/bigdataviewer-vistools, partial update of images in BDV etc). However, some projects have to be finished first and the final functionality has to be made available in ImageJ2-Command developers in an easy manner, especially the BigDataViewer API.

Please feel free to add (simple) use-cases, comments etc as needed.

@fjug @tpietzsch

The issue names no files, tests, or single entry point. Begin by reviewing the four missing capabilities and the referenced RichImg, imglib2-roi, SCIFIOImgPlus, and BigDataViewer APIs; the issue does not define a focused completion criterion beyond making this functionality available to ImageJ2 command developers.

Written by the indexing model from the issue text.