made Mutect2 read position filter default much less stringent

[davidbenjamin]

@takutosato I was checking the filter analysis outputs of every M2 validation and this filter hurts much, much more than it helps, probably because other developments have made it less necessary. Let's essentially turn it off by default.

[xiucz]

Hi,
Sorry to jump in, I find the default value of --min-median-read-position is already 5 in the doc. I am confused why the code's value and the value in the doc are out of sync, it seems that the doc's value was set ahead ?

Have I misunderstood some ? Thanks for your reply.

[davidbenjamin]

@xiucz it was 5; this PR sets it to 1, which is essentially "off." We are gradually converging on a more holistic filtering approach that directly evaluates the possibility of mapping errors, and as we do so these proxies for mapping error outlive their usefulness.