Widen the residual contemplation/vault content-ids to full 256-bit digests and
remove the last default=str / bare-tobytes byte-order gaps (ADR-0244 §2.7 /
ADR-0245 §2.3). Content-addresses must not floor collision resistance or silently
collapse distinct payloads:
- core/contemplation/schema.py: _sha256_16 → _content_digest (full 64-hex);
finding_id + run_id no longer 16-hex truncated (was 2^32 birthday floor).
- core/contemplation/plan_preflight.py: _plan_substrate_hash full digest.
- core/contemplation/miners/articulation_quality.py: drop default=str (as_dict is
JSON-native — ints/str/python-float ratios/None — so a non-serializable field
now fails closed instead of str()-collapsing) + full digest.
- core/physics/holographic_vault.py: _default_mode_id full digest AND explicit
little-endian float64 coercion (was bare .tobytes() — implicit host endianness).
Content-addresses only; no persisted-id migration (old ids stay valid, new ones
widen). Tests assert determinism + content, not id length, so no cascade.
Verified: 334 contemplation/vault/miner tests pass (determinism + content intact).
Full smoke + fast lane run at Phase 5 close.
5c69b74138