Work in progress on binning features in overzoom (#258)

* Factoring out tilestats management from GeoJSON file reading

* Move code around so overzoom can link against parse_layers

* Read the file of bins

* Plumb the bins through to overzoom()

* Some zip code bins to test with

* (Currently non-functional) test of binning

* Starting to spell out the bin matching loop

* Can't flatten points, so don't flatten bins either

* More fleshing out bin traversal

* Bounding box of tile-relative mvt geometry

* Smallest enclosing tile from bbox

* Most of the bin scan

* Add point in polygon check. It crashes.

* Find the matching bins

* GDAL-style bounding boxes have eaten my brain

* Make some features to bin into

* Increment a count as features are found to be within the bins

* Fix longitude wraparound in overzoom bins

* Fix the tests

* Push off attribute copying until after bin assignment

* Carry sum of numeric attributes into the bins

* Also add mean, min, and max

* Add --calculate-feature-index since I keep needing it for testing

* Add an option to accumulate sum/mean/max/min/count of all numeric attrs

* Don't bake in tippecanoe:mean, since we redo it from sum and count

* Forgot to update this test fixture after removing tiled mean

* Update version and changelog

tile-join.cpp

@@ -712,7 +712,7 @@ struct tileset_reader {
 			t.y = parent_tile.y;
 			tv.push_back(std::move(t));
 
-			std::string ret = overzoom(tv, tile.z, tile.x, tile.y, -1, buffer, std::set<std::string>(), false, &next_overzoomed_tiles, false, NULL, false, std::unordered_map<std::string, attribute_op>(), unidecode_data, 0, 0);
+			std::string ret = overzoom(tv, tile.z, tile.x, tile.y, -1, buffer, std::set<std::string>(), false, &next_overzoomed_tiles, false, NULL, false, std::unordered_map<std::string, attribute_op>(), unidecode_data, 0, 0, std::vector<mvt_layer>());
 			return ret;
 		}