Retina module is now parallelized thanks to the TBB library. Speed increase expected on multicore plateforms

added GPU non local means brute force filter + tests

fixed some warnings

(added vim swap files and tag files)

To speed up rebuilds, define the NDK_CCACHE environment or CMake variable
to the path to your ccache binary. When declared, the android.cmake.toolchain
automatically uses CCache when compiling any C/C++ source file. For example:

export NDK_CCACHE=ccache