spec(port): land the remaining CR-CHAT ring batch -- 18 targets (Closes #10579; Refs #8883, #8884)

[dmitrii-f-t27]

Closes #10579

The batch lands the remaining port targets of two rings in one port-wave PR, as specs/policy/pr_budget.t27 (#10067/#10074) asks: one PR per ring batch, never one per file. It supersedes the eighteen one-file PRs opened before that rule. Their spec bodies and seals are unchanged; only the packaging, the ledger rows and the duplicate-body blessing are new.

Targets

CR-CHAT-03 ring (tracker #8883, W3, T-PORT), six targets:

• commit_path_secret_aead_keying_mismatch.t27 (was spec(port): port the commit path secret AEAD keying mismatch guard (Refs #8883) #9875)
• commit_secret_export_collision.t27 (was spec(port): port the commit secret export collision guard (Refs #8883) #9896)
• commit_signature.t27 (was spec(port): port the commit signature forgery guard (Refs #8883) #9904)
• concurrent_add_remove.t27 (was spec(port): port the concurrent add remove reconciler (Refs #8883) #9906)
• psk_external_injection.t27 (was spec(port): port the psk external injection guard (Refs #8883) #9946)
• ratchet_tree_blank_node_depth_guard.t27 (was spec(port): port the ratchet tree blank node depth guard (Refs #8883) #9947)

CR-CHAT-04 ring, wave 2 (tracker #8884), twelve targets:

• padding_nonce_reuse_detection_guard.t27 (was spec(port): port the padding nonce reuse detection guard (Refs #8884) #9709)
• padding_oracle_chosen_ct.t27 (was spec(port): port the chosen-ciphertext padding-oracle guard (Refs #8884) #9738)
• padding_payload_entropy_minimum_guard.t27 (was spec(port): port the padding payload entropy minimum guard (Refs #8884) #9747)
• padding_response_timing_fingerprint_guard.t27 (was spec(port): port the padding response timing fingerprint guard (Refs #8884) #9748)
• padding_timing_side_channel_guard.t27 (was spec(port): port the padding timing side-channel guard (Refs #8884) #9749)
• padding_timing_uniformity_guard.t27 (was spec(port): port the padding timing uniformity guard (Refs #8884) #9750)
• padding_block_alignment_uniformity_guard.t27 (was spec(port): port the padding block alignment uniformity guard (Refs #8884) #9751)
• padding_block_cipher_mode_uniformity_guard.t27 (was spec(port): port the padding block cipher mode uniformity guard (Refs #8884) #9752)
• padding_ciphertext_length_class_balance_guard.t27 (was spec(port): port the padding ciphertext length class balance guard (Refs #8884) #9753)
• padding_key_derivation_uniqueness_guard.t27 (was spec(port): port the padding key derivation uniqueness guard (Refs #8884) #9755)
• padding_key_rotation_uniformity_guard.t27 (was spec(port): port the padding key rotation uniformity guard (Refs #8884) #9756)
• welcome_path_secret_unmasking.t27 (was spec(port): port the welcome path_secret unmasking defense (Refs #8884) #9758)

Why one PR

• One self-consistent duplicate-body bless covers the whole batch. tools/dupe_scan.py --bless records the deliberate self-contained helper copies the ring ports carry, and also the two groups master currently carries unblessed (find, has), so the gate is green on the merge result. Eighteen one-file PRs would each need their own ledger variant.
• About thirty pull-request checks for the batch instead of about five hundred for eighteen one-file PRs -- the queue shape the PR budget policy targets.

Verification, all 18 specs

• t27b test <spec> --blockers: 229 tests total across the batch, 0 failed, 0 blockers.
• t27c test-report <spec>: pass equals tests for every spec, 0 FAIL, 0 vacuous.
• t27c gen-rust <spec> compiles under rustc --edition 2021: 0 &'static str, 0 Vec<const.
• t27c coverage <spec>: Untested 0; ASCII only; module line is line 1.
• t27c seal --verify <spec>: 6 MATCH, 0 mismatch for every spec.
• docs/reports/t27b_expectations.json: 18 rows added, path-sorted, one per spec.
• tools/dupe_scan.py: green on the batch tree.

Refs #8883

Refs #8884

Epic: gHashTag/trios#1812

[gHashTag]

duplicate-bodies will flag this one: padding_key_derivation_uniqueness_guard.t27 defines keys_equal, which is a byte-for-byte copy of ids32_equal in specs/port/trios/crates/trios-chat/common/helpers.t27 (#10161). The fix was verified on the Railway lab: 12/12 tests, 0 vacuous, seal MATCH. Add use port::trios::crates::trios-chat::common::helpers;, delete keys_equal, call ids32_equal, and re-save the seal. Commit a5c13da on the deleted branch sweep/land-9755 has the exact diff. Run python3 tools/dupe_scan.py --like <spec> before pushing; the other new specs in this batch may need the same check.

[dmitrii-f-t27]

Ready for merge.

Battery on the batch tree: 18/18 specs pass (229 tests, 0 failed, 0 vacuous); gen-rust compiles under rustc (0 &'static str, 0 Vec<const); coverage Untested 0; seals 6/6 MATCH per spec; t27b_expectations.json has one path-sorted row per spec; duplicate-bodies green (the bless also records the find/has groups master currently carries unblessed).

The only red checks are master's, not this PR's:

• t27b-native-ratchet: master 436d017 reads 1699/1776 (+3 pass_vacuous, 3 outside; 2204 files). This PR reads 1717/1794 (+3, 3; 2222 files) -- exactly +18 for the added specs, with no new vacuous or outside entries.
• cli-tri: the workflow-file census moved 66 -> 67 because master gained .github/workflows/pr-budget.yml (gate(policy): wire pr_budget.t27 into a pull_request_target gate -- draft the 21st open PR, promote held drafts, exempt queen-* bees #10074). This PR adds no workflow file.

Required checks are green, and the branch merges cleanly against master 8887f17 (local git merge-tree: no conflicts). The batch follows specs/policy/pr_budget.t27: one PR per ring batch. It supersedes the eighteen one-file PRs (#9875, #9896, #9904, #9906, #9946, #9947, #9709, #9738, #9747-#9758), which are closed with linking comments.

[dmitrii-f-t27]

Update after the branch refresh against master: the two native checks (t27b-native-linux, t27b-native-macos) failed this round on one internal t27b test, f32_rounding_and_rem_call_their_ports (cli/t27b/tests/tail.rs:354). Evidence this is a runner flake, not the batch:

• The test is deterministic (fixed inline sources + conformance specs via include_str; none of the batch's 18 specs are read by it).
• Reproduced locally on the exact batch tree: cargo test --release -p t27b --test tail f32_rounding_and_rem_call_their_ports passes, and passes again with the CI seed T27B_DIFF_SEED=3185.
• The latest completed master run of the same workflow is green on both native checks, with the same tree modulo our 18 spec files.

I have no rights to re-run failed jobs on this repository (pull-only), so leaving a note here: re-running the failed jobs of run 39122851031 should clear it. The other red checks (t27b-native-ratchet, Corpus ratchet, Verilog widths, withdrawn-live) are red on master itself — withdrawn-live fails on specs/numeric/paper_claims.t27:198, a file this PR does not touch. Required checks are green and the branch is MERGEABLE.