Better use of location in Computer Vision suggestions

[tiwane]

The iNat team just created a spec for exploring how best to do this, so I’m going to close the topic and voters can get their votes back.