x86 optimization for convolution int8 gemm unified elempack

[nihui]

• xop
• avx2
• avx512vnni
• avxvnni
• opt packA
• opt packB
• avx2/avx512 scatter
• drop old code
• coverage++
• aligned load/store

[codecov-commenter]

Codecov Report

Merging #4881 (5ecf7aa) into master (d38871b) will decrease coverage by 0.18%.
Report is 8 commits behind head on master.
The diff coverage is 95.65%.

@@            Coverage Diff             @@
##           master    #4881      +/-   ##
==========================================
- Coverage   94.90%   94.72%   -0.18%     
==========================================
  Files         776      767       -9     
  Lines      226446   228515    +2069     
==========================================
+ Hits       214908   216468    +1560     
- Misses      11538    12047     +509     

Files Changed	Coverage Δ
src/layer/x86/convolution_im2col_gemm_int8.h	83.89% <ø> (ø)
src/layer/x86/x86_usability.h	100.00% <ø> (ø)
src/layer/x86/convolution_x86.cpp	87.93% <95.65%> (-1.16%)	⬇️

... and 7 files with indirect coverage changes