NVIDIA: fix wrong number of threads

Update for upcoming Graft CNv2 fork

In the cuda backend for monero we start always twice as much threads as needed.
Those threads are than removed after the AES matrix is copied to the shared memory.
Never the less it is the result of an copy past bug.

- start correct number of threads for `monero`

Add console warning on GPU error.

Co-authored-by: Hans Kristian Rosbach <hk-git@circlestorm.org>

Remove files created by macOS

Add missing test for cryptonight_lite

Remove dead code

update version to 2.5.1

update AMD install docu

fix AMD driver 14

- remove link to the AMD APP SDK from a share hoster
- update windows/linux documentation

Fix the fix from #1945. The initial fix produces invalid results.

reduce blocking during metric update

fix broken AMD OpenCL compile

fix access uninitialized value

The AMD compiler for OpenCL shipped with the driver 14XX is broken
and can not compile xmr-stak since the monero v8 changes are introduced.

- workaround a simple compare.
- add new device define `OPENCL_DRIVER_MAJOR`

With #1845 a race condition during the telemetry update is solved.
The problem is that the used mutex is blocking all threads from updating the metrics during
the statistics are calculated.

- introduce a mutex per miner thread

fix #1906

If the currency provided by the user is not known by the miner we access an uninitialized value.

Add default values for member of `coinDescription`.

update version to 2.5.0

NVIDIA: support for multiple CUDA libs

Fix logo on Windows

Allow to ship the miner with multiple cuda backends those depends on different driver versions.
This will allow to support Turing/Volta and old Fermi GPU within one release.

- add support to search for the first working CUDA backend
- add some more messages to support better debugging (if a user has some issues)

update documantion

NVIDIA: tweak `get_reciprocal`

- update tuning guide
- update miner usage

- remove helper array to perform division
- tweak `get_reciprocal`

NVIDIA: rename config option `comp_mode` 

fix right bitshift in `amd_bitalign`

CUDA: fix invalid results

The name `comp_mode` for a memoy load pattern if a bad choosen name.
Therefore I changed it to `mem_mode` which also gives use the possibility
to add new mode later if needed.

- rename `comp_mode` to `mem_mode`
- fix documentation

In the current implementation the bit align is using signed integer which results in pulling in
ones in the case the sign bit is set.

- cast to unsigned integer before using bitshift

If `comp_mode` is false the results on a windows platform will be invalid.
The reason for that is that `ulong4` is in windows 16byte and in linux 32byte.

thx @xmrig for finding and solving the issue

fix #1873

Add Ryo sponsorship message