Unlike GCC and CLANG, Intel Compiler exposes all supported intrinsics,
  whether or not the build options for those features are specified.
  Therefore, we must test #definitions of CPU features when option native/host
  is enabled via `--cpu-baseline` or through env var `CFLAGS` otherwise
  the test will be broken and leads to enable all possible features.

_builtin_popcount is always available, so the compile-time check always
succeeds.

The compile-time test of POPCNT, cpu_popcnt.c produced code that would
execute without error even if the machine didn't support the popcnt
instruction. This patch attempts to use popcnt on random numbers so the
compiler can't substitute the answer at compile time.

  To avoid optimizing it out by the compiler so we make
  sure that the assembler is getting involved.

  - Extending Distutils::CompilerOpt to allow adding extra separate test cases
    related to a certain CPU feature without affecting its availability.

  - Add test cases for reduce intrinsics and AVX512BW mask operations, they can be
    reached through C #defentions NPY_HAVE_AVX512BW_MASK and NPY_HAVE_AVX512F_REDUCE.

  Implement new distutils class `CCompilerOpt`, used for handling
  the CPU/hardware optimization, starting from parsing the
  command arguments, to managing the relationship between the CPU baseline
  and dispatch-able features, also generating the required C headers
  and ending with compiling the sources with proper compiler's flags.

  `CCompilerOpt` mainly used as a helper class for `CCompiler`,
  and doesn't provide any runtime detection for the CPU features,
  instead only focuses on the compiler side, but it generates
  abstract C headers that can be used later for the final
  runtime dispatching process.