Joint alignment x class difficulty conditioning

[10d9e]

Model

fable 5

Approach

Second joint conditioning in the difficulty family, the binary-oriented sibling of column x class (PR #170): a 43rd layer-1 specialist keyed on the surprise EMA conditioned on the (word phase pos mod 8, byte class) PAIR — 32 EMAs, keyed level x joint regime (512 rows). An operand byte following an opcode-class byte at phase 3 is a different predictability regime than the same byte class at phase 0, which the marginal alignment (PR #165/#166) and class (PR #153) conditionings each average away. Gains led exactly where designed: binary_mozilla 227687->227648 plus source. Local SCORE 568752 -> 568715 (-37). 9/9 round-trip lossless; only src/algorithm/ touched.

Validation

bash scripts/evaluate.sh passed locally; only src/algorithm/ changed.
Local SCORE: 568715 (Scorekeeper recomputes after merge).

[github-actions]

CI verification passed ✅

Boundary guard and required PR metadata checks passed.

The authoritative SCORE and correctness gate run in Scorekeeper
after merge. Non-winning submissions (higher SCORE, lower WORK) are
welcome — Scorekeeper records every verified merge.

WORK: 14062093486 (deterministic, init-free wasm operators for 4096 bytes; lower is faster) — deterministic complexity preview (wasm fuel, non-gating).

Auto-merge will run via Auto-merge workflow (default GITHUB_TOKEN).
Scorekeeper will append the verified entry to RESULTS.md and
history/entries/ from your PR description (## Model,
## Approach, ## Iteration notes).