Support struct in map_json

[mielvds]

In continuation of #42, could a polars struct be allowed as input for map_json or maybe a map_struct? When I have a JSON as part of a dataframe, for instance by using df = pl.read_json(file_path), I feel like it's too much encoding/decoding to json.

This would allow something like

df = pl.read_json(file_path)
s = df.get_column("Results")
m = Model()
m.map_json(s, graph="urn:graph:tmp")

[magbak]

This is a good idea! But what if we define instead a mapping of Apache Arrow backed dataframes using facade-x, where this is just the particular mapping applied to structs? Then you could do m.map_df(df) instead. We already have such a mapping for datatypes built. Bit confusing with map() already existing and taking a df, but perhaps that could be map_template. It is maplib so there should be a lot of "map"!

[mielvds]

I agree @magbak and maybe handling json should be left to polars instead of having a map_json in maplib. I leave the design of the API to you, but happy to provide you with feedback from our use cases.