Marble map and clustering

Hi all :-)

First of all thanks a lot for merging the Marble branch (which surely was no
trivial task). I really think it's desireable to become independent of
libkgeomap, and using Marble directly is certainly the way more future-proof
choice.

Still, I think we have some fine-tuning to do, esp. with the clustering. To be
honest, I don't really get which way Johannes implemented it (yet) ... so I
bitch around here without being able to improve it ;-) Take it as constructive
criticism for now ...

The clustering behaves a bit odd. Here's an example from my testing database:

https://imgur.com/a/yQ4AOLi

The first screenshot shows a quite widely north-south extended rectangle. If I
zoom in one step, I get two square rectangles.

As you can see, they are both almost on the same latitude (Actually, the
squares mark the photo's places close enough). So where does this north-south
extension one step above come from? It's not only a bit, the rectangle is
about 200 km tall, whereas the squares of the actual image locations are only
about 20 km apart north-south ...

Also, the actual locations (Konradsreuth and Prague) are not even included in
the first cluster, they are more distant east-west from the shown cluster as
the cluster's width is ...

Surely, there's no rush doing the first marble-based release. But I really
think we should improve the clustering beforehand!

Cheers, Tobias