vg index refactor

[jltsiren]

Changelog Entry

To be copied to the draft changelog by merger:

• Major vg index refactor.
• Always build GBWT from a single graph with a single thread source (VCF, GAM, GAF, paths).
• Index embedded paths as samples instead of contigs with vg index option --paths-as-samples.

Description

This is a major vg index refactor. Many bugs were fixed. Some complex combinations of construction options are no longer supported.

• Threads can no longer be generated in binary format (-H). Use vg gbwt -e to extract them from a GBWT index.
• Always build GBWT from a single input graph. Graphs can be combined with vg index -x --xg-alts.
• Always build GBWT from a single thread source (VCF, GAM, GAF, embedded paths). The indexes can be merged with vg gbwt -m afterwards.
• Providing input XG with -x is deprecated. XG indexes can be used as input graphs.
• GBWT construction from VCF uses less memory, because the graph can be deleted after parsing the VCF for the last contig.
• Index embedded paths as samples instead of contigs with vg index -G -T --paths-as-samples. This supersedes PR this enables haplotype scoring in vg map based on graphs with embedded paths #2862 .