Creating an initial variant list #52

We now have a normalise function to pad out a VCZ with variants from another VCZ that contains the total list of variants (see #30, #42).

How can we create the VCZ with the list of variants? One way would be to use bcftools merge with VCF files that contain different variants (e.g. for different genotype arrays) to get the union of them, then converting to VCZ.

However, this doesn't work in general since bcftools doesn't handle sites with multiple records in a way that keeps the ordering consistent, which is a requirement for Zarr. I've explored this in detail in the repo at https://github.com/tomwhite/vcf-merge-stable, but the gist is that bcftools merge -m none (which won't create any new multiallelic records) can create a variant list that has a different order the input VCFs.

The tool in https://github.com/tomwhite/vcf-merge-stable can create merged VCFs with the desired ordering, as a replacement for bcftools. But we might consider doing the same thing in vczstore. Perhaps something like

vczstore create vcz1 vcz2 ...

to create an empty store with the union of the variants in vcz1, vcz2, etc.