docs(safety): SR-41 for async cross-encoding transcoding; de-stale LS-F-27 (#272)

[avrabe]

What

Closes the #272 campaign's traceability gap (doc-only — no code).

The async cross-encoding string transcoding capability (PRs #273–#285, 44 async_cross_encoding tests) shipped without a requirement covering it: SR-17 links only the sync adapter_safety tests, and LS-F-27's text still said async transcoding was "a larger tracked capability (#272); until it lands, fail loud" — but #272 is done.

Changes

• SR-41 "Async cross-encoding string transcoding" (verified): all 6 directions × {param, result} × {flat, one-level-nested}, both async variants; nested list<u8> deep-copied verbatim; candidate: async adapter may not deep-copy nested string/list pointers in list<string> params (pre-existing, reachability unconfirmed) #281 closed. derives-from SR-17, mitigates LS-F-27, tracked-by #272. Honest verification posture recorded (transcode loops runtime-verified under wasmtime; async-lift integration via the real-adapter *_within_budget tests; full e2e kiln-gated; each increment adversarially Mythos-passed — 4 silent-corruption bugs caught before merge).
• SR-41 traceability entry links a representative set of the async_cross_encoding + integration local-bounds + LS-F-27 gate tests (all verified to exist).
• De-stales LS-F-27: the guard is now the residual fallback for the only remaining unsupported shape (list<list<string>>, #272 follow-up: deeper list nesting (list<list<string>>) — needs a recursive-copy restructure (scoped) #286), allow-set narrowed in lockstep to ≡ the emitter's transcode-set.

Verification

• YAML parses; all 10 SR-41-linked test fns exist; rivet 0 broken cross-refs.
• LS-N gate 57/0/0 (LS-F-27 8 pass — text edit didn't change the ID→test mapping).
• rivet error count +2 = SR-41 joining the pre-existing every-SR-lacks-design-decision/feature backlog (systemic, not new-class, not the required gate).

Non-Tier-5 (safety YAML only) → no Mythos delta-pass.

🤖 Generated with Claude Code

[github-actions]

LS-N verification gate

✅ 57/57 approved LS entries verified

	count
Passed (≥1 test, all green)	57
Failed (≥1 test failure)	0
Missing (no ls_*_NN_* test found)	0

_{Approved loss-scenarios.yaml entries are expected to have a
regression test named ls_<letter>_<num>_* (e.g. LS-A-11 →
ls_a_11_*). The gate runs each prefix via cargo test --lib --no-fail-fast and aggregates pass/fail/missing.}

Failed LS entries

(none)

Missing regression tests

(none)

_{Updated automatically by tools/post_verification_comment.py.
Source of truth: safety/stpa/loss-scenarios.yaml.}