[dmitrii-f-t27]

Refs #8884 (CR-CHAT-04 ring tracker). Parent epic: gHashTag/trios#1812 (W2, own lane, T-PORT). Eighteenth W2 target delivered: the uniformity rules for padding length class transitions — batch ceiling (4096), minimum floor (8), per-record from/to class ceilings (< 4), strictly-increasing u64 timestamps, no-self-transition, then the chi-squared ceiling (15.0) over destination-class counts — D5 verdict carrying idx / got / bound / class / got_f / bound_f / prev / current for every variant, original rule order, all ten PLCT tests transcribed plus chi-squared helper and rule-order tests (12 total, 0 vacuous). Original accumulates usize counts and casts to f64 in chi_squared_flat; t27b rejects the int-to-float cast, so counts accumulate in f64 from the start — identical arithmetic for integer counts (documented in the header). The 4097-record ceiling fixture builds in a loop (batch rule fires before any record). t27b 12/12 native; t27c 12/12; gen-rust compiles under rustc (0 errors, 0 &'static str, 0 Vec<const); coverage Untested 0; seal all hashes MATCH; ledger row included (pass, non-pass untouched). Dedup checked. Sequencing: one-line ledger rebase if the earlier lane PRs land first — I take it on request.

[dmitrii-f-t27]

Merge request for the gHashTag merger: steward READY (required all SUCCESS — validate, check-linked-issue, parse-ratchet, t27b-native-linux, t27b-native-macos; the red ones are non-required per Q16 and red on master too). Ledger row included; one-line rebase if the earlier lane PRs land first — I take it on request.

[gHashTag]

Carried unchanged, with its original author, as one commit of #10097 (tracking #10096). That PR lands all 25 green CR-CHAT-04/07 ports at once, because each single PR kept conflicting on the ledger rows before its CI could finish. Closing this one so its queued jobs free the Actions queue.