Fix: Skip connections have separate 1x1 convolutions from the residual output.

[basveeling]

After some correspondence with the authors it turns out that the skip connections are parameterized by a 1x1 convolution, separate from the one that goes into the add block. This should be an easy fix but could potentially explain poor training results before.

[jfsantos]

All this means is that each skip_out should be the output of another Conv1D layer instead of being the same as x, right? I'll submit a PR soon.