[cuda] Support for cuda 11

[simogasp]

Cuda 11 has dropped the support for sm_30 and sm_32 architecture.
https://docs.nvidia.com/cuda/cuda-toolkit-release-notes/index.html#deprecated-features
Also, it deprecates other older architectures:

Support for Kepler sm_30 and sm_32 architecture based products is dropped.
Support for the following compute capabilities are deprecated in the CUDA Toolkit:

sm_35 (Kepler)

sm_37 (Kepler)

sm_50 (Maxwell)

To reflect this, the script for choosing the CC should be updated as follows:

  set(CUDA_MIN_CC 20)
  set(CUDA_MAX_CC 75)
  if(CUDA_VERSION_MAJOR GREATER_EQUAL 11) #new
    set(CUDA_MIN_CC 35)                   #new 
  elseif(CUDA_VERSION_MAJOR GREATER_EQUAL 10)
    set(CUDA_MIN_CC 30)
  elseif(CUDA_VERSION_MAJOR GREATER_EQUAL 9)
    set(CUDA_MIN_CC 30)
    set(CUDA_MAX_CC 72)

@griwodz Am I correct?

[simogasp]

Ah, it also introduces CC 8.0, shall we add it too?
In that case

  set(CUDA_MAX_CC 80)                   #new 
  if(CUDA_VERSION_MAJOR GREATER_EQUAL 11) #new
    set(CUDA_MIN_CC 35)                   #new 
  elseif(CUDA_VERSION_MAJOR GREATER_EQUAL 10)
    set(CUDA_MIN_CC 30)
    set(CUDA_MAX_CC 75)                   #new 

[simogasp]

It seems to be working with cuda 11 https://travis-ci.org/github/alicevision/popsift/jobs/711458057

[simogasp]

Here are the changes 568a4c9