t27b-native-ratchet red on master: re-bless the ledger for 263 unlisted passing specs #10195

Boundary

docs/reports/t27b_expectations.json only (the t27b per-spec ledger). Rows are written by tri t27b ratchet --bless --accept-new, never by hand. No workflow, compiler or spec change.

Scenarios

The t27b-native-ratchet check is red on master and so on every pull request. Its verdict on master run 39094812421 (commit fdeb1db): tri t27b ratchet: red -- run fdeb1db16 vs ledger a9ad52584: UNEXPECTED FAILURE 0, UNEXPECTED PASS 0, UNLISTED 263, STALE 0, OVER CAP 0, BAD REASON 0, MOVED 0, UNJUDGED 0, VACUITY MEASURED 0.

The ledger was last blessed at a9ad525. Since then 263 specs reached the corpus (specs/t27f/, specs/skills/t27-t27f.t27, specs/claw/, the CR-CHAT and gen_w port batches, specs/igla/coder/, specs/nn/). The reference passes all of them, but the ledger does not name them. Of the 263, 217 are t27b pass, 3 are pass_vacuous, and 43 are t27b blocked on a construct t27b does not lower yet. t27b-native-ratchet is not a required check, but it makes every PR look red and hides real regressions.

Requirements

• Bless the ledger from the native arm64 CI run of master (artifact t27b-native-run of run 39094812421) on the Railway lab, with --accept-new. The 43 new non-pass specs raise max_not_pass 33 -> 76, which the tool accounts for as exactly the new specs (33 already named + 43 new; 0 named entries left the run).
• Every other count stays: UNEXPECTED FAILURE, UNEXPECTED PASS, STALE, MOVED, BAD REASON and OVER CAP are all 0. No existing row changes.

Criteria

• tri t27b ratchet --run run.json against the blessed ledger prints green with UNLISTED 0.
• The diff only appends 263 entries and moves the header (source commit/ref, max_not_pass 33 -> 76).
• CI t27b-native-ratchet is green on the PR.