Put ./. genotype when there is no power to identify a variant

[mfoll]

At the moment we put 0/0 when QVAL<50, but this can be due to the absence of mutation but also absence of power to detect a mutation. This should now be easy now that we implemented #133.

We need to have a parameter to choose the allelic fraction we are trying to identify:

• For germline variants we should use the same parameter use to classify somatic/germline status (20% at the moment).
• For somatic mutation this will be a new parameter.

[tdelhomme]

@nalcala nextflow part seems to be ok, I've also added lines in the R script to recover the values, keeping your previous synthax (afmin and sigma).