PostGIS Functions over-costed

[pramsey]

As discussed:

​https://lists.osgeo.org/pipermail/postgis-devel/2022-September/029818.html

In general, the high (and arbitrary) costs assigned to some PostGIS functions are flipping plans into less desirable states. On the repeatable case discussed, cutting the high cost in half allowed the preferred plan to prevail. In general we want to keep our function costs high, since compared to pretty much all PgSQL core functions we do have some extremely high computation needs, and want plans that reflect that.

[Paul Ramsey <pramsey@…>]

In 44e6da9/git:

Reduce "high" and "medium" costs by a factor of two.
Follows this conversation:
​https://lists.osgeo.org/pipermail/postgis-devel/2022-September/029818.html
Very high costs are flipping some plans badly.
Want to retain relatively large costs to promote parallel
plans and other things that are "good" for geospatial,
but walk the line carefully.

Closes #5259