Support different output types for `summary` endpoint

[zacharyDez]

Currently, the summary endpoint returns a List of the rows for each hex_id. We added the optional support for geometries, but a user wanting to generate a GeoDataframe must still convert the Geojson feature for each row into a Shapely point.

During code review, @vincentsarago suggested we implement a behavior similar to TiPG, where the output format of the response for summary depends on an output_format parameter. When a user requests a geometry to be attached to the response, we could return a FeatureCollection, which can be read directly by geopandas and other tools.

[zacharyDez]

@bpstewar: I'm questioning the necessity for this one. I did some initial design and thought about it. Adding different output formats would complicate the API interface and the underlying implementation. The added overhead for a user to get valid GeoJSON is a single line:

df['geometry'] = df['geometry'].apply(lambda geom: shape(geom))

And even if we were to return a GeoJSON Feature Collection, an end-user would need to perform some data wrangling to get it in the correct format for their analysis (DataFrame).

[zacharyDez]

Based on a discussion with @bpstewar and @andresfchamorro, we let users process the initial summary response to match their needs. The needs are varied, and developers can transform them to fit their needs.