Refine download smoke fragmentation gate

[1613358894]

Summary

• refine bounded download-smoke FASTA fragmentation classification so complete multi-replicon outputs with one dominant chromosome-scale record are not automatically blocked
• keep WGS/scaffold/contig header keyword signals fail-closed under the default fragmentation profile
• add focused coverage for the GCF_000011805.1-style complete multi-record FASTA shape observed in bounded server validation

Rationale

A bounded one-row server smoke downloaded a Complete Genome/reference assembly with three FASTA records and no WGS/scaffold/contig header keywords. The old local inspection treated any non-90%-dominant multi-record FASTA as multi_record_fragmented, which over-blocked an otherwise high-quality complete multi-replicon genome. This change keeps scaffold/contig/WGS-like outputs blocked while allowing a small number of records with one dominant chromosome-scale record to remain ready for review/installability smoke.

Validation

• python scripts/check_workspace_hygiene.py
• python scripts/check_docs_hygiene.py
• python -m pytest tests/test_cli_download_smoke.py -q --basetemp .pytest_tmp_fragmentation_gate2 -p no:cacheprovider
• python -m pytest tests/test_docs_consistency.py tests/test_docs_hygiene_script.py -q --basetemp .pytest_tmp_fragmentation_gate_docs2 -p no:cacheprovider
• git diff --check

No live provider/API lookup, broad download, external bioinformatics tool, tag, release, or asset upload was run for this PR.