Hello twesigomwedavid ,
in my opinion "pros and cons" for a particular variant caller isn't useful in the case of CYP2D6. The challenge in variant calling for this gene is, that there is a highly homolog gene (CYP2D7) next to it. That makes it difficult for the aligner to determine the origin of the reads. Furthermore it is well know, that there are recombination between the two genes.
I have no best strategy for calling variants in CYP2D6 from NGS for you. I did this with plain old sanger sequencing before, where I can amplify the CYP2D6 locus to make sure not to get the sequence of the CYP2D7 locus.