Optimize loading performance of name mapper

[steffen]

The name mapper is used mainly in the background when read or write well known texts (wkt).

At the first use of the name mapper, the mapping information is loaded from the huge (1.4MB) dictionary file 'NameMapper.csv'. For further use the name mapper is then cached. The loading is time consuming. Currently in a standard environment the loading takes about 500ms.

Optimize the name mapper file loading procedure. Reading the file at once into the memory and then parse it from there would gain performance due less file IO access.