[Proj] New proposal of JNI bindings for Proj.4

[Martin Desruisseaux]

Hello all

An example of Proj.4 usage in Java has been documented there:

http://www.geoapi.org/geoapi-proj4/index.html

Only the first example is relevant to Proj.4. The second example is an 
abstraction level on top of it.

As a side note, GeoAPI provides a "conformance" module which can be used for 
testing arbitrary implementations of GeoAPI interfaces. Every points used for 
those tests come from documented sources like EPSG or national mapping agencies 
(see for example 
http://www.geoapi.org/geoapi-conformance/apidocs/org/opengis/test/referencing/ParameterizedTransformTest.html#testMercator1SP%28%29). 
I have run those tests on Proj.4 and got the expected results with centimetric 
precision, except:

   * Cassini-Soldner for which I had to relax the tolerance threshold to 2 metres
     (note: this is the same in Geotoolkit.org, which is derived from Proj.4 code
     anyway)

   * EPSG:31300 (Belge 1972 / Belge Lambert 72) which seems to be 745 km away
     in Easting compared to the example point given by the EPSG database. I don't
     know if the error come from my misunderstanding of Proj.4 (I get the expected
     result with Geotoolkit.org). Steps to verify in the EPSG database:
       - Go to http://www.epsg-registry.org/
       - Click on "Retrieve by code", enter 9803
       - Select the " Lambert Conic Conformal (2SP Belgium)" method
       - Click on "Example" on the bottom right side.

     Regards,

         Martin