fix: quote-aware VCF/BCF header key=value parser

[jmg421]

Problem

scanBcfHeader() fails to parse structured VCF/BCF header lines whose quoted field values contain commas or = signs. Two cascading regex bugs in .bcfHeaderAsSimpleList():

1. Comma split — the lookahead ',(?=Description=".*?")' uses lazy .*? which still breaks when the quoted value contains = (e.g. Description="IDs [VRS version=2.0.1]")
2. Key=value split — (?<=[ID|Number|Type|Description])= uses [...] (character class, matches single chars I,D,|,…) instead of alternation, so it splits on = signs inside quoted values too

Reproducing examples from the tracker:

# VariantAnnotation#89
INFO=<ID=VRS_Allele_IDs,Number=R,Type=String,Description="The computed identifiers [VRS version=2.0.1;VRS-Python version=2.1.1]">

# VariantAnnotation#80
GATKCommandLine=<ID=SelectVariants,CommandLineOptions="analysis_type=SelectVariants input_file=[] showFullBamList=false">

Fix

Replace both strsplit calls with .splitKeyVals(), a character-by-character quote-aware scanner that:

• Splits on commas only outside double-quoted strings
• Splits each field on the first = only (key names never contain =)

The helper is hoisted to package scope (was previously inlined) so unit tests and downstream packages can access it via Rsamtools:::.splitKeyVals.

Changes

• R/methods-BcfFile.R — replace regex split with .splitKeyVals()
• inst/unitTests/test_BcfFile.R — 3 regression tests covering both issues + commas-in-quotes
• NEWS — changelog entry under v2.18

References

• possible problem parsing INFO description field VariantAnnotation#89
• readVcf doesn't handle quoted = in hapmap_exome_chr22.vcf.gz correctly VariantAnnotation#80