x86 optimization for convolution int8 gemm unified elempack #699

	name: linux-x64-cpu-gcc-musl
	on:
	push:
	branches: [master]
	paths:
	- '.github/workflows/linux-x64-cpu-gcc-musl.yml'
	- 'toolchains/host-c.gcc.toolchain.cmake'
	- 'CMakeLists.txt'
	- 'cmake/**'
	- 'src/*'
	- 'src/layer/*'
	- 'src/layer/x86/**'
	- 'tests/**'
	- 'tools/**'
	- '!tools/pnnx/**'
	- 'examples/**'
	pull_request:
	branches: [master]
	paths:
	- '.github/workflows/linux-x64-cpu-gcc-musl.yml'
	- 'toolchains/host-c.gcc.toolchain.cmake'
	- 'CMakeLists.txt'
	- 'cmake/**'
	- 'src/*'
	- 'src/layer/*'
	- 'src/layer/x86/**'
	- 'tests/**'
	- 'tools/**'
	- '!tools/pnnx/**'
	- 'examples/**'
	concurrency:
	group: linux-x64-cpu-gcc-musl-${{ github.ref }}
	cancel-in-progress: true
	permissions:
	contents: read

	jobs:
	linux-gcc-musl:
	runs-on: ubuntu-latest
	steps:
	- uses: jirutka/setup-alpine@v1
	with:
	packages: >
	cmake
	clang
	clang-dev
	make
	gcc
	g++
	libc-dev
	linux-headers

	- uses: actions/checkout@v3
	- name: build
	shell: alpine.sh {0}
	run: \|
	mkdir build && cd build
	cmake -DNCNN_BUILD_TESTS=ON ..
	cmake --build . -j 2
	- name: test
	shell: alpine.sh {0}
	run: cd build && ctest --output-on-failure -j 2
	- name: build-shared
	run: \|
	mkdir build-shared && cd build-shared
	cmake -DNCNN_SHARED_LIB=ON ..
	cmake --build . -j 2

Provide feedback