Revisit current "DepthMap" implementation

[NiklasRentzCAU]

Currently we use the DepthMap to pre-calculate detail levels (out of bounds, minimal details, full details) for "regions" before rendering, and using that information in subsequent svg rendering calls.
I think we do not need to pre-calculate this at all and can therefore remove the DepthMap entirely, in exchange for simply doing an out-of-bounds and relative size check during the rendering (computationally inexpensive, done so per default in Sprotty's examples). Furthremore, some pre-rendering-checks such as calculating the absolute scale of elements can then be either done in some other, fast pre-rendering step or also on the fly during rendering itself, stored in properties of the graph for persistence.

This would remove this clutter inconsistent with Sprotty and will also solve other issues related to the DepthMap #77.