"flat" polygon has non-zero area according to ST_Area

[strk]

Spin-off of #5105, ST_Area can report non-zero area for completely flat polygons. Example:

POLYGON((
  29.262792863298348 71.22115103790775,
  29.26598031986849  71.22202978558047,
  29.275379947735576 71.22044935739267,

  29.29461024331857  71.22741507590429,

  29.275379947735576 71.22044935739267,
  29.26598031986849  71.22202978558047,
  29.262792863298348 71.22115103790775
))

Note how the polygon ring is going forward and backward onto the exact same vertices.

[strk]

Automated test for the above is in

​https://gitlab.com/postgis/postgis/-/merge_requests/71

[strk]

The code is ptarray_signed_area and references ​https://en.wikipedia.org/wiki/Shoelace_formula with a comment saying it should return 0 for "flat" polygons, which is not happening.

[strk]

GEOS suffers from the same bug, although PostGIS does NOT use GEOS for this: ​https://github.com/libgeos/geos/issues/581

[pramsey]

Per ​https://github.com/libgeos/geos/issues/581 I'm closing this. Do enough arithmetic on FP and you get some drift. Maybe figure out your tolerance and round?