Feature .id, .geometry, .properties, .type properties

[sgillies]

Lately I'm feeling that, for example, ftr.properties['foo'] is cleaner syntax than ftr['properties']['foo'] and that we could surface all the standard GeoJSON items as Python properties in addition to the old item getter style. So, ftr.id == ftr['id'] would be True.

We could get this by requiring the geojson module or could roll our own dict-based classes.

[geowurster]

@sgillies Definitely cleaner syntax and more Pythonic but I wonder if it will cause confusion when some GeoJSON libraries have a Feature() class and others don't?

[sgillies]

We could update the Python geo protocol while we're at it. Would head off some confusion and help clear up the remainder.

[geowurster]

Yeah that and recommending people check for __geo_interface__() rather than assuming that an object that appears to be GeoJSON actually is, like our potential Feature() and Geometry() classes.

I sketched out a bit of this ticket in #1.