Add precision parameter to ST_AsEWKT

[Algunenano]

While working in the shortest representation for doubles I've noticed an approach used multiple times to print a EWKT (i.e. with the SRID) up to a certain precision:

SELECT ST_AsEWKT(ST_SnapToGrid(g, 0.01))

This approach is wrong because ST_SnapToGrid does not guarantee that the coordinates will fall into that precision; and it really can't, since there might not be a floating point representation for that number.

Instead, what I think should be available is

SELECT ST_AsEWKT(g, 2)

And leave ST_SnapToGrid to its real use case, gridding (which is why it does stuff like removing duplicates).

[Algunenano]

PR implementing it: ​https://github.com/postgis/postgis/pull/575

[Raúl Marín <git@…>]

In f0822be/git:

Add precision parameter to ST_AsEWKT

Closes #4698
Closes ​https://github.com/postgis/postgis/pull/575