Improvement try to add feature threading

[kalwalt]

Enhance nft marker creation with pthreads

First step

I will test if we can improve the speed up creation of the markers, for this reason will try to distrubute the calculation in multiple threads.
This PR provide the first step: convert the existing code from C to C++, as we will try to use std::mutex and std::thread.
It also address issue #2 and #3.

[kalwalt]

For now i have only converted the code to C++, so we can use std::chrono features and modern threading. I have to fix also these warnings:

stderr: /src/emscripten/markerCreator.cpp:241:19: warning: ISO C++11 does not allow conversion from string literal to 'char *' [-Wwritable-strings]
            usage("Marker Generator");
                  ^
/src/emscripten/markerCreator.cpp:244:26: warning: ISO C++11 does not allow conversion from string literal to 'char *' [-Wwritable-strings]
        char *filename = "tempFilename";
                         ^
/src/emscripten/markerCreator.cpp:463:45: warning: ISO C++11 does not allow conversion from string literal to 'char *' [-Wwritable-strings]
            if (ar2SaveFeatureSet(filename, "fset", featureSet) < 0)
                                            ^
3 warnings generated.

[kalwalt]

Now i used strdup and it tooks 188.294593 seconds to create the nft markers instead without strdup it tooks 195.639931 seconds, (using the pinball.jpg image)

[kalwalt]

With the last commit i'm testing pthreads features but i donìt see any improvements in the execution time, why this?

[kalwalt]

I think that my implementation does simply run the computation in separate thread and do not distibute the computation in multiple threads. this should be done inside the ar2GenFeatureMap function. This involve to develop a separate function whit mutlithreading, I think i will do this in another PR.