Generating invalid multipolygon

[zebdelk]

Found a record failing to index because the multipolygon we make isn't valid: http://localhost/registry/metadata/collection/unknown/3ef068c5-f359-43ae-b304-6eddf35dee3e/parsed

error message from geojsonlint.com is Line 1: Polygons and MultiPolygons should follow the right-hand rule

We need to be VERY careful fixing this, so we don't break other multipolygons!

two potential sources of bug:

1. incorrectly identified as crossing antimeridian when it doesn’t - shouldn’t be a multipolygon at all
2. does cross antimeridian, should be a multipolygon, but somehow order of points is wrong

this example should be fixed:

• [ ] generates a valid multipolygon or polygon
• [ ] generates the correct covering polygon(s) for this record (see potential sources of bug)
• [ ] the test cases are complete enough to make sure we don’t break other multipolygons that are correct and working, and this case is added to that set

A reminder - we create multipolygons when bbox crosses the antimeridian in compliance with this recommendation: https://tools.ietf.org/html/rfc7946#section-3.1.9 This is because some (large) bboxes are indexed by ES incorrectly otherwise, and other solutions may cause issues with other tech that handles antimeridian or points outside the -180 to -180 range differently.

[dneufeldcu]

Java Topology Suite probably has code to handle/check this

[mcquinne]

Need to find a better example of this situation... the record indicated here could be fixed by a metadata manager, but if this is a real bug we need to find a clear example.