Preparation for somatic CNV

[ericblanc20]

The somatic CNV pull requests is becoming a monster.

It is safer to break it into manageable items if possible. The first step is to create new steps that are required by one or several somatic CNV tools. In particular:

• guess_sex: cnvkit and ascat both require knowledge of the patient's sex, but what to do when it's missing from the sample sheet? (which is unfortunately very often). This provides a simple-minded way to provide some value, based on autosomes & sex chromosome coverage. This might be later become as complex as we want (see for example purple).
• germline_variant_calling: cnvkit, purecn, control-free & ascat require vcd files containing both gremlin & somatic variants in order to improve segmentation & assess purity & LOH. Mutect2 can be run with arguments to keep germline variants, but the resulting somatic calls are very different from those without germline ones. This is not acceptable, so we need to find another solution. Unfortunately, the variant_calling step has been developed with rare diseases in mind, and cannot just deal with one single file.

This issue and the accompanying PR aim at including a few steps & routine which are needed for the new CNV computation.