duplicate-bodies: share the 12 PTX emitter helpers between the forward and backward PTX specs #10947

The duplicate-bodies workflow ("Duplicate Body Ratchet") fails on master since #10855 added specs/igla/coder/hslm_bwd_ptx.t27: twelve PTX emitter helpers are byte-identical copies of the ones in specs/igla/coder/ternary_gemm_ptx.t27, and none of the groups is in tools/duplicate_bodies_baseline.txt. Every open PR inherits the red. Supersedes the shape-refused #10930 and the duplicate-bodies part of #10935.

Boundary

• specs/igla/coder/ptx_emit.t27 (new shared helper spec)
• specs/igla/coder/ternary_gemm_ptx.t27
• specs/igla/coder/hslm_bwd_ptx.t27
• their seals under .trinity/seals/ and bootstrap/gen/rust/igla/coder/ternary_gemm_ptx.rs and hslm_bwd_ptx.rs (regenerated, never hand-edited)
• No Rust edits. Out of bounds: the scan and orphan reds of HSLM release close-out: clear master's inherited reds (scan, duplicate-bodies, orphans) so #10836 lands #10935.

User Scenarios

• A contributor opens a PR and sees duplicate-bodies red with the names at_text, code_at, count_text, emit_epilogue, emit_lds, emit_mma_step, emit_store, ins, put_quad, put_s, put_u, weight, none of which the PR touched.
• A maintainer changes the PTX text writer once and both the forward and the backward kernel pick it up.

Requirements

• FR-001: The twelve helpers MUST live in one spec, specs/igla/coder/ptx_emit.t27, and both PTX specs MUST use it instead of keeping a copy.
• FR-002: The emitted PTX text MUST NOT change: every existing test of both specs still passes, 0 vacuous.
• FR-003: The new spec MUST carry its own tests (L4) and be sealed; the two importers MUST be resealed and their gen-rust output regenerated by t27c, not by hand.
• FR-004: The duplicate-body ledger scan MUST report no new and no grown group, with no new ledger row.

Success Criteria

• gh run list --repo gHashTag/t27 --workflow "Duplicate Body Ratchet" --branch master -L 1 --json conclusion --jq '.[0].conclusion' prints success on the master tip after the fix merges.
• The ledger scan (tools/dupe_scan.py) prints ok: no new duplicate body, and none of the known groups grew.
• t27c test-report on ptx_emit.t27, ternary_gemm_ptx.t27 and hslm_bwd_ptx.t27 on the Railway lab: all pass, 0 vacuous; t27c seal --verify prints all hashes MATCH for the three.