Swap pilot: five chat guards run t27c gen-rust output behind adapters; original tests unchanged #1895

Part of #1812. Uses the baseline of #1825 (PR #1827).

Goal

Prove the swap end to end on the 5 trios-chat ports that are already merged in gHashTag/t27 and whose t27c gen-rust output compiles. Record the cost per file (minutes and adapter lines) so the remaining waves are estimated from measurements.

Files

Files this PR touches

Owner work done by dmitrii-f-t27 in PR (linked below); not a bee task, please do not shape or dispatch it.

• crates/trios-chat/rings/CR-CHAT-01/src/key_package_expiry_guard.rs
• crates/trios-chat/rings/CR-CHAT-01/src/t27gen/key_package_expiry_guard.rs
• crates/trios-chat/rings/CR-CHAT-02/src/epoch_rollover_wraparound_guard.rs
• crates/trios-chat/rings/CR-CHAT-02/src/t27gen/epoch_rollover_wraparound_guard.rs
• crates/trios-chat/rings/CR-CHAT-03/src/ratchet_tree_resolution_guard.rs
• crates/trios-chat/rings/CR-CHAT-03/src/t27gen/ratchet_tree_resolution_guard.rs
• crates/trios-chat/rings/CR-CHAT-03/src/unmerged_leaves_bound_guard.rs
• crates/trios-chat/rings/CR-CHAT-03/src/t27gen/unmerged_leaves_bound_guard.rs
• crates/trios-chat/rings/CR-CHAT-07/src/cover_decoy_ratio_governor.rs
• crates/trios-chat/rings/CR-CHAT-07/src/t27gen/cover_decoy_ratio_governor.rs
• crates/trios-chat/rings/CR-CHAT-01/Cargo.toml, CR-CHAT-02/Cargo.toml, CR-CHAT-03/Cargo.toml, CR-CHAT-07/Cargo.toml (only a [lints.rust] check-cfg entry)
• crates/trios-chat/T27_PIN

User Scenarios

• Given the five original guard files at trios 8b229e9489ee, when their logic is replaced by t27c gen-rust output behind an adapter, then every original test of the four rings passes unchanged and the frozen e2e, falsifier and vector outputs are identical.
• Given a vendored generated file, when one comparison in it is flipped, then at least one original test fails; when it is regenerated, all pass again.

Requirements

• FR-001: each rings/<R>/src/t27gen/<stem>.rs MUST be byte-identical to t27c gen-rust output at the t27 SHA recorded in T27_PIN, and its sha256 MUST equal the seal's gen_hash_rust.
• FR-002: each rings/<R>/src/<stem>.rs MUST keep the original pub consts (original types), error enums with derives and #[non_exhaustive], docs and the original #[cfg(test)] module byte for byte.
• FR-003: generated items MUST stay private (mod t27gen, never pub mod).
• FR-004: no new compiler warnings outside mod t27gen.

Success Criteria

1. For each of trios-chat-cr-chat-01, -02, -03, -07: cargo test -p <pkg> -- --list 2>/dev/null | grep ': test$' | sort | diff - crates/trios-chat/golden/test-names/<pkg>.txt prints nothing, and cargo test -p <pkg> reports the baseline count with 0 failed.
2. For each swapped file: diff <(git show 8b229e9489ee:crates/trios-chat/rings/<R>/src/<stem>.rs | sed -n '/^#.cfg(test)./,$p') <(sed -n '/^#.cfg(test)./,$p' crates/trios-chat/rings/<R>/src/<stem>.rs) prints nothing.
3. $T27C gen-rust <spec> | cmp - crates/trios-chat/rings/<R>/src/t27gen/<stem>.rs is silent with t27c built at the pinned SHA; shasum -a 256 of each vendored file equals the seal's gen_hash_rust.
4. cargo test -p trios-chat --test t27_pin passes (5 rows); cargo test -p trios-chat --test r_chat_guard gives 3/3.
5. e2e_chat_25 PASS list, falsifier_runner stdout and exit code, and golden_vectors output are identical to golden/.
6. Mutation proof for all 5 files: flipping one comparison in each t27gen/ file makes at least 1 original test fail; regenerating restores all tests, and cmp against fresh gen-rust output is silent. Counts in the PR.
7. Minutes and adapter lines per file are recorded in the PR and copied to the epic.
8. The required checks no-js-check, Test, Constitutional Enforcement and guard are green on the latest SHA.

Not in this issue

• The t27 issues 7634, 8366 and 8125 stay open although their PRs merged; closing them needs triage rights on gHashTag/t27.
• The ratchet_tree_resolution_guard spec asserts only ok / not ok. Making its tests assert the exact variant is a W1 follow-up in gHashTag/t27.

Dmitrii Fedorov (@dmitrii-f-t27)