noise: filter activities spatially

[tschlenther]

Addresses #3548

This is just a draft with comment representing my idea how to do this

[tschlenther]

I resolved the issue #3548 by excluding activities in the damage calculation that are outside the grid, or more precisely by NOT mapping them to the closest grid cell.
I tested that via the NoiseDashboardTests, which use a mini Kelheim scenario. There, in local tests, I set the grid (via ShpOptions) to a small area which covers a highway. There are many noise immissions but no activities. See the red area in this screenshot as a reference (blue dots are home activities at hour 0)

Before this PR, the result would look like this

where you see that edge cells pointing towards Munich (south) etc have a high damage value.
After this PR, the result looks like this:

where you see no damage at all (there is no activity).