Batch PRs #166, #170, #171, #157, #159

[Psy-Fer]

Batch merge of five small, independent PRs from @BenjaminDEMAILLE. Each is preserved as its own commit(s) with original authorship; the batch is split by commit so it reads PR by PR.

Closes #166, #170, #171, #157, #159.

What's in it

PR	Commits	What
#166	a10b29b, 86ffe98	--runThreadN 1 ran on every logical core. The rayon pool was only configured above 1, and skipping build_global leaves rayon's default of one worker per core.
#170	164b297, 7d7b367	Drop Transcript::read_seq, a public field nothing reads — a full read copy per finalised alignment.
#171	a4d6226	test/bench_ab.sh: interleaved A/B harness that samples CPU idle before and after every run and drops rounds it could not measure honestly, rather than folding them into the median.
#157	d181ac3, 4f7b6fe, 1a343b2, caec836	--soloCellReadStats CB writes Solo.out/<feature>/CellReads.stats, plus --genomeChrSetMitochondrial.
#159	ae575c7	--runMode soloCellFiltering <raw dir> <output prefix> — cell-call an existing raw matrix without aligning.

Changes made while merging

None. All five applied cleanly and nothing needed correcting. The only manual step was a CHANGELOG.md conflict between #157 and #159, which both add an entry at the same anchor; both entries are kept, resolved into ae575c7.

Two things checked rather than assumed, since they are the claims that carry risk:

• refactor: drop Transcript::read_seq, a public field nothing reads #170 removes a public field. The compiler proves nothing read it, and the 99 deleted lines are the field plus its initialiser in test fixtures — 56 of them in io/sam.rs alone, one line per Transcript literal. No production code path changes.
• fix: --runThreadN 1 ran on every core, not on one #166 now calls build_global unconditionally. run_thread_n is NonZeroUsize, so there is no zero case to worry about.

Validation

All five are output-neutral by construction — a thread-pool fix, a dead-field removal, a shell script, and two features reached only via new flags. Verified rather than assumed:

Byte-identical to current main on the default paths, records compared with headers excluded (the @PG CL: line necessarily differs by output prefix):

	records differing from main
SE 10k yeast	0
PE 10k yeast	0
STARsolo, 10x mouse chr19	0

Solo Gene raw matrix.mtx, barcodes.tsv and features.tsv are all byte-identical to current main on the same run.

Against STAR 2.7.11b, unchanged from the recorded baselines:

• SE 10k yeast: 8788/8926 (98.5%)
• PE 10k yeast: 8390 both-mapped, 0 half-mapped, per-mate position agreement 16605/16780 (99.0%)

Gate: 593 tests pass, cargo clippy --all-targets clean, cargo fmt --check clean.

# SE
./target/release/rustar-aligner --runMode alignReads --genomeDir $DATA/indices_rustar \
  --readFilesIn $DATA/reads/ERR12389696_sub_1_10k.fastq.gz --readFilesCommand zcat \
  --outFileNamePrefix out_se_ --outSAMtype SAM --runThreadN 1
python3 test/compare_sam.py --rustar-aligner out_se_Aligned.out.sam --star $DATA/star_10k_/Aligned.out.sam

# PE
python3 test/compare_pe.py out_pe_Aligned.out.sam $DATA/star_10k_pe_/Aligned.out.sam