DONE: selected k8 singleton B-frame cap (seed 29; PR #1358)

[1337isnot1337]

DONE — PR #1358, head 381d4a48b3cab8c204d437627b4e76cfc0c17164.

On the selected PR1094 k8 word, all 147 unchanged B-gate frames with dimension four have common future-frame caps of dimension four (122 gates) or five (25 gates). Any legal one-gate 4D-to-6D replacement must lie inside its cap, so the complete 22,785 gate/superspace domain is excluded. The exact per-gate caps and source pins are in the PR package. This is a scoped finite obstruction; no new kappa, multi-gate optimality, or full multiplication theorem is claimed.

Reproduce:
python3 -I -B research/complex-k8-singleton-frame-cap/verify.py --out /tmp/complex-k8-singleton-frame-cap.json

Expected: PASS_NO_ADDITIONAL_4_TO_6_SINGLETON_B_FRAME gates=147 candidate_space=22785 max_cap_dimension=5 chronologically_feasible=0.

The selected result remains unchanged. The verifier and cached certificate replay matched; staged whitespace checks passed. Full selected cold replay and hosted CI were not rerun.