Circular chromosomes in the VCF header

[nh13]

Similar to #403, I'd like to formalize the use of topology (especially for circular contigs) so that variants may span the origin of a circular contig without having to split into multiple records, use SV notation, etc.

[jmarshall]

I feel like at the time we planned to extend this to VCF, but I can't find any record of that now.

I guess the natural VCF translation of the @SQ TP field would be an optional reserved attribute in §1.4.7 Contig field format, something along the lines of

topology: Molecular topology, as per SAM's @SQ TP header field. Either linear or circular.