called prior to transformations now by default

also set pe/simd for non-DW for faster results

so that we can run it with the correct parallelization method in the test suite

[Test] ConvInputGen+VVAU node test to check behaviour of depthwise convolution in hls layers and update VVAU node