spec(port): port the cover-traffic payload length distribution guard (Refs #8887)

[dmitrii-f-t27]

Refs #8887 (CR-CHAT-07 ring tracker). Parent epic: gHashTag/trios#1812 (W2, own lane, T-PORT). Thirteenth CR-CHAT-07 ring target delivered: the length-distribution rules (R-CHAT-10) — batch ceiling (128), per-batch repeated-id scan ([8]u8), class floor (2) / ceiling (16), observation floor (10), chi-squared ceiling (50.0; exact boundary passes in CTPL-09, the rule strictly greater) — D5 verdict with every payload field, original rule order, all ten CTPL tests transcribed plus helper and rule-order tests (12 total, 0 vacuous). The chi-squared statistic arrives in the record exactly as the original carries it (caller-computed over the histograms — the record's own field); BTreeSet → visited-scan. 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.