Tighten scope of SubGridContainer

[ckittl]

I found a corner case, where our definition of SubGridContainer's scope get challenged:

Let us have a very detailed grid model with properly modeled switches at the high voltage side of the transformer. As of now, we assume, that the seem of the subnets is as depicted in the lower part of the picture. However, I feel more like it should be the case as depicted in the middle of the figure. There are two reasons, why I think it would be better:

1. The switches more or less belong to the concept of the transformer, which is also situated in subnet 2.
2. If I would have the same grid, but with a lesser degree of detail (especially without switches), node A would as well be the seem between the two subgrids.

@johanneshiry @sensarmad I highly would appreciate your thoughts on this, so that we can decide on the next steps to address this Issue.

[ckittl]

As of discussion we take the A | A' solution, as in future scenarios, the interest to operate on the switches would be more likely on the LV side.

[ckittl]

The following tasks have to be addressed:

• Adapt the sanity checks, when setting up SubGridContainers
• Determination of predominant voltage level: Exclude not only the high voltage node of the transformer, but all nodes "above", that are connected by switches
• Improve the creation of the sub grid topology graph