Lower strictness of criteria for 'expected support levels'

Our algorithm for defining variants which we expect RNA support of is currently quite strict. When applying to real samples we consistently see it toss away many variants where support is observed. We should drop the default `min_alt_supporting_reads` to 1 supporting read in RNA