feat(igla): Wave Loop 896 — module-scope [611][2]^6 Pt non-power-of-two outer-dimension array-of-struct variable from call with indexed signed writes #1855

Next mechanical rung of the packed-vector array-of-struct ladder.

• Variant A: module-scope [611][2]^6 Pt variable initialized from a pure make_grid(0) call, with indexed signed field writes and assert_eq read-back inside a bench block.
• Expected: 39,104 elements, 1,251,328-bit packed vector, ~1.194 MiBit.
• Continues the post-1-MiBit boundary stress test.
• Zero compiler / reference-model / FROZEN_HASH changes expected.

Refs the Wave Loop autopilot plan.

User Scenarios

• Given a .t27 spec declaring a module-scope [611][2]^6 Pt array-of-struct variable initialized from a pure make_grid(0) call with indexed signed field writes and assert_eq read-back inside a bench block, when the spec is compiled and executed by t27c, then the bench passes reporting 39,104 elements, a 1,251,328-bit packed vector, and ~1.194 MiBit.

Requirements

• FR-001: The compiler MUST accept module-scope non-power-of-two outer-dimension array-of-struct variables with ^6 packed-vector annotation.
• FR-002: The compiler MUST correctly initialize such variables from pure make_grid(0) calls.
• FR-003: The compiler MUST support indexed signed field writes and assert_eq read-back inside bench blocks for these variables.
• FR-004: The resulting packed vector MUST have exactly 39,104 elements and 1,251,328 bits (~1.194 MiBit) with no change to FROZEN_HASH.

Success Criteria

• t27c <spec-file> --bench exits 0 and prints "39,104 elements, 1,251,328-bit packed vector, ~1.194 MiBit"
• cli/t27b/bench/bench.py runs the new bench and reports 1 test passed
• .trinity/seals/Power.json remains unchanged (verified by git diff --exit-code .trinity/seals/Power.json)

Boundary

• specs/tri/collections/array.t27
• specs/tools/trinity/tri/bench.t27
• specs/memory/graph/scope.t27
• specs/tools/tri/loop.t27
• specs/fpga/power.t27
• cli/t27b/bench/bench.py
• .trinity/seals/Power.json