Handle rows with variance = 0
We need to handle rows without variance as we did in coexpnetviz before trying to calculate the correlation https://gitlab.psb.ugent.be/deep_genome/coexpnetviz/blob/master/coexpnetviz/_algorithm.py#L149
In CoExpNetViz we simply dropped such rows, I wonder how morph handles missing genes in expression matrices as this results in missing genes.