Properly detect and enable vectorization on ARM

!320 (merged) has the side effect of breaking detection of SVE vectorization support and enablement of SVE in the compiler. My patch should properly fix the underlying problem.

py-cpuinfo is supported on ARM64 and can be used to detect Neon and SVE. However, there was indeed a bug here -- Neon is identified as asimd in /proc/cpuinfo, so we should check for asimd instead of neon.

While -march=native was not supported by Clang before 15, -mcpu=native is supported by GCC 6+ and Clang 7+. Let's use that instead of not adding a flag at all -- otherwise SVE support is not enabled in the compiler even if the hardware supports it.

pystencils/backends/simd_instruction_sets.py

@@ -43,8 +43,7 @@ def get_supported_instruction_sets():
         return _cache.copy()
     if 'PYSTENCILS_SIMD' in os.environ:
         return os.environ['PYSTENCILS_SIMD'].split(',')
-    if (platform.system() == 'Darwin' or platform.system() == 'Linux') and platform.machine() == 'arm64':
-        # not supported by cpuinfo
+    if platform.system() == 'Darwin' and platform.machine() == 'arm64':  # not supported by cpuinfo
         return ['neon']
     elif platform.system() == 'Linux' and platform.machine().startswith('riscv'):  # not supported by cpuinfo
         libc = CDLL('libc.so.6')
@@ -72,7 +71,7 @@ def get_supported_instruction_sets():
     required_sse_flags = {'sse', 'sse2', 'ssse3', 'sse4_1', 'sse4_2'}
     required_avx_flags = {'avx', 'avx2'}
     required_avx512_flags = {'avx512f'}
-    required_neon_flags = {'neon'}
+    required_neon_flags = {'asimd'}
     required_sve_flags = {'sve'}
     flags = set(get_cpu_info()['flags'])
     if flags.issuperset(required_sse_flags):