perf: implement Conv1D with im2col and GEMM #2618

Job	Run time
build-and-push-docker-images	23m 0s
build-and-test-cpp-aarch64 (openblas)	4m 7s
build-and-test-cpp-x86_64 (ubuntu-20.04, mkl)	5m 32s
build-python-wheels (ubuntu-20.04, auto64)	26m 48s
build-and-test-cpp-x86_64 (ubuntu-20.04, dnnl)	4m 51s
build-python-wheels (macos-11, auto64)	35m 36s
check-python-style	10s
build-and-test-cpp-x86_64 (macos-11, mkl)	10m 29s
build-python-wheels (windows-2019, auto64)	44m 41s
build-and-test-cpp-x86_64 (macos-11, dnnl)	9m 37s
build-python-wheels (ubuntu-20.04, aarch64)	1h 11m 19s
build-python-wheels (macos-11, arm64)	12m 33s
test-python-wheels (ubuntu-20.04)	9m 32s
build-and-deploy-docs	31s
test-python-wheels (macos-11)	4m 49s
test-python-wheels (windows-2019)	4m 32s
publish-python-wheels-on-pypi	0s
	4h 28m 7s

Provide feedback