feat(dada): add --failed-uniques diagnostic (issue #60)

[cjfields]

Closes #60.

What

Adds --failed-uniques <FILE> to dada, dada-pseudo, and dada-pooled. It emits the unique sequences that failed to denoise (final p-value < omega_c, map == null — R's $map == NA) as a tidy long-format TSV:

sequence	sample	reads
TACGAAGG…AACA	soil_A	7
TACGGAGG…AATC	soil_A	3

One row per failed unique per sample it appears in, with the reads it cost. This answers "what failed to denoise and how many reads did it cost?" without hand-joining the dada map to the derep uniques. Motivated by benjjneb/dada2#1899.

Semantics follow the subcommand

Subcommand	"Failed" decided	reads is
dada, dada-pseudo	per sample (map == null; for pseudo, in the round-2 pass)	the unique's in-sample abundance
dada-pooled	globally (result.map == null on the merged unique table)	the failed merged unique's read count in that sample (one row per sample it appears in)

Rows are sorted by sample / descending reads / sequence, so output is deterministic regardless of concurrent denoising order.

Note on the issue's caveat

The issue flagged that a pooled per-sample map null could mean either a denoising failure or a cluster with zero reads in that sample. That second case is unreachable: a unique present in a sample always carries ≥1 read into its cluster, so a per-sample map == null always means a genuine denoising failure. The TSV reports only real failures.

Changes

• new module src/failed_uniques.rs (Row + write_tsv, deterministic sort)
• denoise_and_serialize returns (json, Vec<Row>) gated by a collect_failed bool
• collectors wired into single/multi dada, both dada-pseudo round-2 closures, and the dada-pooled output loop
• integration test dada_failed_uniques_matches_map_nulls (TSV row count ≡ JSON map null count)
• docs/diagnostics.md: new --failed-uniques section, cross-linked to kdist-calibrate --from-dada

Testing

• full suite (60 unit + 13 integration) and clippy pass
• verified end-to-end on all four modes; single-sample row count (60) matches the JSON map null count exactly

🤖 Generated with Claude Code