Do both HG002 and NA12878 (aka HG001)

[adamnovak]

Right now the test script only uses NA12878. It would be good to also use HG002 as a test sample.

The challenge here is generating the haplotype graphs needed to simulate the reads, since this sample isn't in the 1kg VCFs; we may need to add a second vg construct step just to get the graphs to simulate reads from.

We also would need duplicate mapeval and calleval steps, because each only handles a single sample.