Cor.ind values may be wrong

[antongrau]

The correlation values per individual do sometimes have a strange behaviour. We get complex numbers as results. This is clearly unwanted behaviour.
We need a reference example for these values.
Perhaps they should be disabled until trusted???