thread affinity for non cpu backends

- check if affinity vale before calling `thd_setaffinity`

amd: use 64bit target

This is a follow up of #43 and use the some mechanism to set the thread affinity for
non cpu backends correct.

- use cpu affinity workflow for nvidia and amd
- cpu: move messages of thread spawning before thread creation

- remove 32bit target value
- use always 64bit target for amd backend

remove double blob copy

- remove double copied memory

prefere AMD OpenCL over NVIDIA, Intel

Implement pool-controlled nonce allocation

fix #56

Search for AMD OpenCL inlcude path and libaries to prioritize the AMD implementation over NVIDIA and Intel.

Proper removal of gpu_thread_num

fix illegal memory access

fix CUDA 9 shuffle warning

add NVIDIA Volta support

fix assert in AMD backend

remove restricted pointer

fix bug announced in #52

If the miner is compiled in debug mode the assert condition results in a compile time error.

add `sm_70` if CUDA 9+ is avalable

use `__shffl_snyc` if CUDA 9+ is avalable

add order_fix for double hash

Make hwloc check capabilities and stop it from leaking mem

Re-add UAC manifest and WIN_UAC cmake option

Affinity fix version C

Make sure all singletons are set to null and make env a global ptr

remove the gpu_thread_num safety

Keep consistency between class and struct