spec(port): port the padding class transition audit (Refs #8884)

[dmitrii-f-t27]

Refs #8884 (CR-CHAT-04 ring tracker). Parent epic: gHashTag/trios#1812 (W2, own lane, T-PORT). Twelfth W2 target delivered: the transition audit over a sequence of class choices — every class canonical (256/1024/4096/16384), any window of ten non-decreasing with at most window_size/2 class changes, every class run but the last at least PCTA_MIN_PER_CLASS = 2 messages — D5 verdict, original rule order, all ten PCTA tests transcribed plus helper, rule-order and under-window tests (13 total, 0 vacuous). The windows(10) sliding scan becomes an explicit window loop; the sentinel 4294967295 is not a canonical class, so an over-long table trips the canonical rule rather than reads as a choice. t27b 13/13 native; t27c 13/13; gen-rust compiles under rustc (0 errors, 0 &'static str, 0 Vec<const); coverage Untested 0; seal all hashes MATCH; ledger row included (pass, 28/28 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; 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.