empty raw WKB policy about sf HOT 2 CLOSED

What happens is that Manifold "Normalizes" the geometry for export, and this geometry is a degenerate polygon with no y-extent: http://www.georeference.org/forum/t133518.3

So, arguably the query should do the filtering on the Manifold side given the possibility of these degenerate outputs,, or insert a valid NULL geometry with the cast (to plug a hole in Manifold's coercion) until their WKB output is fixed.

Still, sf could have a nicer message on error, or invoke a "missing geom" policy.

from sf.

I'm not so in favor of accepting invalid WKB: where to stop? I'll solve it with the nicer error message.

from sf.