[cschanhniem]

Implements J10 from NEXT_100_PR_MAP: end-to-end pipeline smoke test.

What

• tests/test_pipeline_dry_run_e2e.py: 63 tests across 3 classes
• TestDryRunPipelineSteps: 9 tests, one per pipeline module (FASTA export, JSON-LD context, adapter stub, schema manifest, release manifest, NCL link, changelog entry, docs coverage, stale doc detector)
• TestDryRunSafetyGates: 4 tests, verifies dry-lab gates cannot be bypassed
• TestDryRunConsistencyChecks: 7 tests, cross-schema consistency (prefix uniqueness, shared format vocab, toy sequence validity)
• All candidates use TOY- prefix; no external calls; no disk I/O
• BASELINE 8221→8284

Why

Without an E2E smoke test, each schema module is tested in isolation but nobody checks that they chain together. This test verifies the key modules introduced in today's Phase I/J work (FAE-, JLC-, ADS-, SEM-, RMF-, NCL-, changelog, checks) can be assembled into a coherent pipeline run with consistent inputs and outputs.

Disconfirming pass

• real candidate ID (non-TOY prefix) rejected under dry_lab_only
• public_release context rejected under dry_lab_only
• wet-lab constraint=False rejected in JSON-LD
• dry_lab_only=False rejected in adapter stub
• Toy sequences validated against VALID_AMINO_ACIDS and length bounds