add missing __syncthreads to edgesHysteresisLocalKernel

replace std::swap with own code

* out of border access in edgesHysteresisLocalKernel
* incorrect usage of atomicAdd

- new reduce implementation (with kepler optimizations)
- saturate_cast via asm command
- video SIMD instructions in element operations
- float arithmetics instead of double
- new deviceSupports function

2) device layer utility

added convert functions to BruteForceMatcher_GPU
other minor fixes