feat(region): UCSC<->Ensembl chromosome-name aliasing for -r

[balwierz]

Summary

A -r chr1:… query against a file that names the contig 1 (or vice versa) silently returned zero rows — the classic UCSC-vs-Ensembl naming footgun a bioinformatician hits constantly. vv now retries with the counterpart name when the queried chromosome isn't present but its alias is, noting the swap on stderr.

Scoped deliberately (per the constraints): only human and mouse standard chromosomes are aliased — autosomes 1–22 (covers human 1-22 and mouse 1-19), X, Y, and the mitochondrion chrM ↔ MT (never bare M). Scaffolds, patches, and alt-contigs are never remapped, so a genuine typo (e.g. chr99) still errors/empties exactly as before.

chrom_alias() + a templated resolve_chrom() / resolve_region_chroms() resolve each region window's chromosome against the file's own contig list, applied where each source knows its contigs:

• tabix (.vcf.gz/.bed.gz/.gff.gz/.tsv.gz) via tbx_seqnames
• BCF via bcf_hdr_name2id
• BAM --pileup via bam_name2id
• LociSSD v4 via rank_to_name

Generic-Parquet and bigWig region paths (which don't carry a cheap contig list) are a noted follow-up.

$ vv variants.vcf.gz -r chr1:1000-2000     # file uses '1'
vv: variants.vcf.gz: region chromosome 'chr1' not found; using '1' (UCSC/Ensembl naming)
… rows …

Verification

• New Ensembl-named fixture tiny.ens.bed.gz (contigs 1, MT).
• Tests: 1↔chr1 on LociSSD v4 / tabix VCF / BCF (equal row counts); the chrM→MT special case; the reverse chr1→1; and that a non-standard chr99 is not remapped (0 rows). Suite 370.
• Man page + CHANGELOG updated.

🤖 Generated with Claude Code