spec(port): share the duplicated CR-CHAT helpers blessed by #10098 #10159

Boundary

Touches only .t27 specs under specs/port/trios/ (the CR-CHAT-03/04/07 ring specs, three trios lib specs, two new shared modules), the 29 matching .trinity/seals/*.json files and tools/duplicate_bodies_baseline.txt. No Rust, no gen/, no tooling code.

Background: PR #10098 folded 66 port PRs into one merge and blessed 12 duplicate-body groups into the ledger. Eight of them are plain copies of the same small helper (burst_ids_equal x13, emission_ids_equal x10, burst_id_zero x7, bytes_equal x5, all_zero_bytes x4, dest_zero x3, sched_sqrt x3, approx_sqrt x2). The other four (id_seen_before, obs_seen_before, dest_count, member_contains) read fields of ring-local record structs or use different integer widths, so they cannot be shared without generics or a type change.

Scenarios

1. A reader fixes a bug in a byte-equality helper: today they must find and patch 13 copies; after this change they patch one module.
2. A new CR-CHAT ring needs a 16-byte id comparison: it uses the shared module instead of pasting another copy.
3. CI runs the duplicate-bodies check: it is green and the ledger is eight lines shorter.

Requirements

1. Add specs/port/trios/crates/trios-chat/common/helpers.t27 (ids16_equal, ids32_equal, zero_id16, bytes_equal, all_zero_bytes, approx_sqrt) and specs/port/trios/common/newton_sqrt.t27, each with tests that execute asserts.
2. Replace every copy site with a use of the shared module; keep behaviour identical.
3. Reseal the touched specs and re-bless the ledger on the Railway lab.
4. State plainly in the PR that the four record-bound groups remain.

Criteria

• Every touched spec passes the t27c test report with 0 failures and 0 vacuous passes.
• tools/check_seal_coverage.py reports OK.
• tools/dupe_scan.py reports no new duplicate body and no known group grown, after the re-bless.
• The ledger loses eight lines.