feat: add bamqc subcommand for generic genomic BAM/CRAM QC

[BenjaminDEMAILLE]

Closes #128.

Adds rustqc bamqc — generic genomic QC in one streaming pass, no GTF required:

rustqc bamqc sample.bam --outdir results/
rustqc bamqc sample.cram --reference genome.fa --outdir results/

What it reports

Coverage depth and breadth (genome-wide and per contig), GC content, insert size, mapping quality, read lengths and base composition — the metrics side of Qualimap's bamqc mode.

Output names and layout match Qualimap, so MultiQC's Qualimap BamQC module picks them up unchanged:

genome_results.txt
raw_data_qualimapReport/coverage_histogram.txt
raw_data_qualimapReport/genome_fraction_coverage.txt
raw_data_qualimapReport/mapped_reads_gc-content_distribution.txt
raw_data_qualimapReport/mapped_reads_nucleotide_content.txt
raw_data_qualimapReport/insert_size_histogram.txt

Coverage is exact, not sampled

Each read contributes +1 over every aligned block (M/=/X/D); N skips leave a gap. Since input is coordinate-sorted, depth is finalised as the file streams past, so memory scales with pile-up depth rather than genome size — no per-base array.

Secondary alignments are excluded from coverage. Duplicates are included by default (Qualimap's default); --skip-duplicated excludes them.

Validation

Cross-checked against samtools depth on tests/data/test.bam, both modes, and pinned in integration tests:

Mode	RustQC	samtools
default	mean 0.6038X, 12.16% ≥ 1X, 4.78% ≥ 5X	samtools depth -a -J -g DUP → identical
--skip-duplicated	mean 0.4325X, 11.55% ≥ 1X	samtools depth -a -J → identical

(The two differ by exactly the 137 duplicate-flagged reads × 50 bp in the test file.)

Scope

This is the metrics subset, not all of Qualimap bamqc. Not included: the HTML report, per-window coverage tracks, clipping profile, homopolymer indel breakdown, PCR-bias/chimerism estimates, and coverage restricted to a target BED (--feature-file). The docs page states this explicitly. Happy to extend if you'd like any of those in the first cut — and the offer in the issue to validate against real Qualimap output on real data would be well worth taking up, since my validation here is against samtools depth rather than Qualimap itself.

Note this overlaps in machinery with #18 (single-pass CRAM QC): the depth tracker here is what a mosdepth-equivalent would build on.

• cargo test — 204 lib + 20 integration tests pass
• cargo fmt --check, cargo clippy -- -D warnings clean
• New docs page docs/bamqc, AGENTS.md updated

🤖 Generated with Claude Code