* Crunch down memory required per polygon joint

* *Actually* reduce the size of the structure

* Update version and changelog

When running "make test" in a pristine working copy, the "test" target
calls the "raw-tiles-test" target, which tries to execute ./tile-join.
But that binary has not been built, yet, at this point. The target fails
with an error.

Add the required binaries as dependencies to the test targets in the
Makefile. This makes sure that they are built before they are attempted
to be executed.

* Start of externalizing polygon shard detection

* Completely untested external quicksort

* Add unit test for external quicksort

* Remember to clean up temporary files

* Sort and scan the vertices

* Bring over more vertex logic

* Make nodes from vertices

* Checkpoint on switching over to global shared nodes

* Do the thing

* Revert unintended change to coalesced linestring behavior

* Take shared nodes into account in early simplification

* Let it do more sorting in memory

* Fix overnoding of collinear linestrings

* Remove duplicate nodes, since only duplicate vertices now matter

* Remember to delete temporary files

* Fix out of bounds memory access below, apparently

* Fix the actual undefined behavior

* Still running out of memory in one case. Find out where.

* Forgot the conditional

* Try again to make it not run out of memory

* Update version and changelog

* Clip away entire features by bbox. Avoid unnecessary recompression.

* Move parent tile decoding in tile-join out of overzoom proper

* An ever-growing cache of parent tiles

* Limit the size of the cache

* Remove the current reader *before* checking if we can run the queue

* Clean up

* Add missing #include

* Add comment

* When the tile-join cache fills up, evict the least recently used

* Fix microsecond math

* Factoring out tile-join's cache for testing

* Add unit tests for tile-join cache

* Update changelog and version

…elt#149)

* Only detect longitude wraparound within each ring.

Between rings it's OK to jump around from one side of the world
to the other.

* Add test

* Update changelog

…felt#161)

* Special-case a longitude wraparound of exactly 360°

* Update version and changelog

* Make tile-join distrust source tilesets' metadata maxzoom and minzoom

* 16 bits is enough for tile numbers

* Revert "16 bits is enough for tile numbers"

This reverts commit 71a0c4e.

* Check what child tiles each overzoomed tile will have

and don't queue further overzooming of empty tiles

* Clean up naming; use std::move to avoid copying large arrays