Now everything is prepared for the pull request.

Some interface was refined (most notably, the method for returning
Hessian was removed and the method for getting gradient was added as
optional to base Solver::Function class) and basic code for
setters/getters was added. Now is the time for the real work on an
algorithm.

Implementation is based on
http://www.cs.cmu.edu/~quake-papers/painless-conjugate-gradient.pdf

So far we only have basic interface and empty test. But it compiles at
least.

Streamlined mexext detection, with removal of trailing whitespace/newlines handled by execute_process

first parameter of createERFilterNM1/createERFilterNM2 is now mandatory. changed the sample program to use the new prototypes

renamed getDefaultClassifierNM{1,2}() to loadClassifierNM{1,2}() and added "const std::string& filename" parameter (mandatory).

functions getDefaultClassifierNM1() and getDefaultClassifierNM2() allow to implicitly load the default classifiers when creating a ERFilter object

fix thin-prism coefficients at 0's in the stereo calibration function unless the flag CALIB_THIN_PRISM_MODEL is explicitly set

I didn't make the status part of the version variables here, because
letters don't play too well with version comparison algorithms
(especially with CMake's, which only supports numbers).

This will allow us to distinguish between regular releases, prereleases
and development versions.

Aligned strings in CMakeList

Modified the check on plantuml extension

Some fixes

Moved find_host_program for plantuml script