Selecting a point (instance) and selecting a pixel is conceptually different when it shouldn't

[Wumpf]

Today images are logged as blobs and points as list of components. But as a user this distinction isn't all that meaningful, especially from Python where in both cases one logs numpy arrays.

Hovering & selecting an image:

-> We clearly hover a pixel, but select the whole image

Hovering & selecting a point:

-> We clearly hover a point and select a point. However, we don't get to see any information of the object to which the user logged all these points.

UI wise we'd like to have "both in both cases": We want to get information about the object holding the data (in one case the image, in the other the point cloud object) but also want to be aware of the fact that we selected a single pixel and a single point!

[Wumpf]

we decided that internally pixels will be identified via instance ids. That implies that images conceptually a collection of pixel instances together with some meta data! See #1756