I was wondering about the theoretical differences and use-case applications for the 2 different GATK CNV calling pipelines: germline and somatic

I know that the somatic should be more sensitive than the germline but let's consider 2 real use-case scenario:

1. potentially pathogenic CNV discover in a particular disease (let's say a rare disease): would it be better to first generate a PoN and then call cases against it with the somatic pipeline? Or generate a cohort and use case-mode against it with the germline one?
2. analysis of all ploidy changes and CNV in an immortalized cell-line (so theoretically a lot): somatic or germline?

Also the somatic one produces very great plots that the germline does not. Why is that?

Thank you so much in advance for any help!