ddc_pipeline get 'best' CUDA device + add SAMPLE_CLOCK_STAR in when 'disk' or...
- Pure CUDA/CPP based DDC -> removed cupy based DDC
- ddc_pipeline get 'best' CUDA device + add SAMPLE_CLOCK_STAR in when 'disk' or 'dummy' capturing
- VDIF pipeline allow to run on 1 NUMA node
- cpp: ddc benchmark info only in DEBUG
- Revised cmake project