dtype issue in set_pois #15
Description
I still think there's an issue in the python/C exchange when calling set_pois that pops up in certain cases, though I cant tell exactly when or why. The example in the example notebook works fine, but if I try to swap out the geosnap data in favor of the DC boundary from osmnx (i.e. oxdc = ox.geocode_to_gdf('washington d.c.'), then the set_pois call crashes out. There are also some weird differences between the networks that dont make sense to me; i'll post a short reproducible gist.
It's not my highest priority since the POI functionality is less important in most of my own use-cases, but it's still a genuine bug afaict. I think it has something to do with integer indices from the osm_ids being too large (i.e. somewhere in the C code, int needs to be changed to long or something), but i'm really not sure