feat(igla): Wave Loop 882 — module-scope [583][2]^6 Pt non-power-of-two outer-dimension array-of-struct variable from call with indexed signed writes

[gHashTag]

Continue the mechanical packed-vector array-of-struct ladder.

• OUTER = 583, MID_IDX = 291
• Elements = 583 × 2⁶ = 37,312
• Packed vector = 37,312 × 32 = 1,193,984 bits ≈ 1.139 MiBit
• Variant A from W881 cooperation plan: non-power-of-two outer-dimension [583][2]^6 Pt module-scope variable from call with indexed signed writes.
• Zero compiler / reference-model / FROZEN_HASH changes expected for witness.

Acceptance criteria:

• Generator scripts/gen_w882.py with copy hazard fixed before first run.
• Spec specs/scratch/w882_bench_module_583x2p6_aos_var_call_write.t27 parses.
• t27c icarus-lowerable, icarus-simulate, icarus-cocotb, and seal --save PASS.
• Integration test added to bootstrap/tests/icarus_lowerable.rs.
• Full cargo test --release --test icarus_lowerable passes at 342/0.
• bootstrap/stage0/FROZEN_HASH unchanged.
• Closeout report + W883 cooperation plan + tracker updates.
• Commit with Closes #{issue}, push branch, open PR.

[gHashTag]

Closing — nothing left to land once this is measured against current master

Checked with the GitHub API against master at 10e3d5e71a970351d5d1c63f0dd88b77210dbffc. This PR touches four paths, and none of them can still land:

Path	Status on master
scripts/gen_w882.py	already there, byte-identical (blob 77c9f845f41488a3f5960155a5db5b27f9434492), landed via commit 0e2442540c80 — "Wave Loop 882 … (Closes #1812)"
bootstrap/tests/icarus_lowerable.rs	master already carries accepts_w882_bench_module_583x2p6_aos_var_call_write among its 352 accepts_w* tests; this hunk adds nothing new. The +1890 count is an artifact of an ~848-commit-old merge base. No test is removed by this close
specs/scratch/w882_bench_module_583x2p6_aos_var_call_write.t27	specs/scratch/ is gitignored (.gitignore:113) and does not exist on master
.trinity/seals/scratch_w882_bench_module_583x2p6_aos_var_call_write.json	gitignored (.gitignore:120); master carries zero scratch_* seals, so re-adding this one would red the seal gate

Worth recording rather than losing: this PR never contained a W882 narrative record — no docs/reports/FPGA_LOOP_CLOSEOUT_W882_*.md, no .claude/plans/wave-loop-883.md, no .trinity/experience.md entry. master's experience log runs W898 → W884 and then jumps to W777, so W881–W883 have no entry there. Closing this PR discards nothing that ever existed; the gap is separate, and only the W883 half of it is recoverable (from #1815).

The branch wave-loop-882 is left in place, so the scratch witness stays reachable if it is ever needed.

Series close-out: #1810, #1813, #1842, #1844, #1847, #1850, #1852, #1854, #1856 and #1858 are all closed for this reason. #1815 was the only PR in the series with unique durable content and was rebuilt on top of master.