* stop using 'CUDA' prefix
  * add explanation

  * add feature of Fp16 on GPU (cudev)
  * add test
  * leave template function as unimplemented to raise error

it might be called from _OutputArray::setTo

- 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

added getParams method to VideoWriter_GPU

update gpu tests