Fixed GIS tests with Oracle 23c.

[smithdc1]

My attempt at a patch to allow GIS tests to pass with Oracle 23c.

The change to LayerMapping is due to Oracle 23c changing "MULTI" geometries containing one value to their single type.

create table LAYERMAP_COUNTY_TEST
(
    ID       NUMBER(11),
    NAME     NVARCHAR2(25),
    STATE_ID NUMBER(11),
    MPOLY    MDSYS.SDO_GEOMETRY
);

INSERT INTO "LAYERMAP_COUNTY_TEST" ("NAME",
                               "STATE_ID",
                               "MPOLY")
VALUES ('Galveston', 3, SDO_GEOMETRY('MULTIPOLYGON (((30 20, 45 40, 10 40, 30 20)))',4269));

INSERT INTO "LAYERMAP_COUNTY_TEST" ("NAME",
                               "STATE_ID",
                               "MPOLY")
VALUES ('Hawaii', 4, SDO_GEOMETRY('MULTIPOLYGON (((30 20, 45 40, 10 40, 30 20)),
((15 5, 40 10, 10 20, 5 10, 15 5)))',4269));

select SDO_UTIL.TO_WKTGEOMETRY("LAYERMAP_COUNTY_TEST"."MPOLY")
 from LAYERMAP_COUNTY_TEST;

"POLYGON ((30.0 20.0, 45.0 40.0, 10.0 40.0, 30.0 20.0))"
"MULTIPOLYGON (((30.0 20.0, 45.0 40.0, 10.0 40.0, 30.0 20.0)), ((15.0 5.0, 40.0 10.0, 10.0 20.0, 5.0 10.0, 15.0 5.0)))"

[smithdc1]

Marking this PR as draft.

This is the edit to make the test pass, but it seems wrong to me. This new point is Dallas (entering the point at https://www.geometrymapper.com/), before it was Forney. Note above would suggest this change is wrong.

Thoughts welcome 😄

[felixxm]

Maybe it's related with undefined ordering 🤔

[smithdc1]

Taking another look at this morning. Just leading a small note here for now to say that I've just realised the supports_tolerance_parameter feature (and therefore this test) is only enabled on Oracle.

[smithdc1]

I'm suspecting these test failures are due to version differences in GDAL/GEOS.

There's equals_exact() which accepts a tolerance. That would be useful here, but what's tricky here is that we're comparing two sets. I'm not sure the Points can be ordered in a deterministic(?) way first.