This commit adds a new model class which can be used by any app to
interact with a broadcast area. A broadcast area is one or more polygons
representing geographical areas.
It also adds some models that make browsing collections of these areas
more straightforward. So the hierarchy looks like:
> **BroadcastAreaLibraries*
> Contains multiple libraries of broadcast area
> > **BroadcastAreaLibrary**
> > A collection of geographic areas, all of the same type, for example
> > counties or electoral wards
> > **BroadcastArea**
> > Contains one or more shapes that make up an area, for example
> > England
> > > **BroadcastArea.polygons[n]**
> > > A single shape, for example the Isle of Wight or Lindisfarne
> > > > **BroadcastArea.polygons[n][o]**
> > > > A single coordinate along a polygons
The classes support iteration, so all the areas in a library can be
looped over, for example if `countries` is an instance of
`BroadcastAreaLibrary` you can do:
```python
for country in countries:
print(country.name)
```
The `BroadcastAreaLibraries` class also provides some useful methods for
quickly getting the polygons for an area or areas, for example to
render them on a map. So if `libraries` is an instance of
`BroadcastAreaLibraries` you can do:
```python
libraries.get_polygons_for_areas_long_lat('england', 'wales')
```
This will give polygons for the Welsh mainland, the Isle of Wight,
Anglesey, etc.
The models load data from GeoJSON files, which is an open standard for
serialising geographic data. I’ve added a few example files taken from
http://geoportal.statistics.gov.uk to show how it works.
