duplicate place names

[cldellow]

Even after handling label-nodes-in-relations for places, we have duplicates:

This is probably a data bug on the OSM side. For example, Mount Vernon:

• node: https://www.openstreetmap.org/node/141064238#map=19/48.42005/-122.32642
• relation: https://www.openstreetmap.org/relation/237865#map=12/48.4230/-122.3106

The node is not linked to the relation :(

Ideas to fix:

• fix the OSM data - this sounds interesting, but is a big task and would engage the OSM bot policies, so not an immediate solution
• de-dupe in Tilemaker -- this would require a new feature in Tilemaker, to let some Lua code run on the per-tile output (it would also be useful for addressing Redundant housenumbers in addresses layer shortbread-tiles/shortbread-tilemaker#3, I think)
• de-dupe in Lua -- maybe based on Wikidata ID, or name attribute?
  • simplest thing that could work: write a file with the wikidata ID, check its existence in relation phase
  • better would be a real database...but then we have a dependency on a real database. Maybe we could use SQLite if we sharded on ID?