128-column difficulty conditioning

[10d9e]

Model

fable 5

Approach

Applies the phase-widening lesson from the alignment sweep (PR #166) to the line-column difficulty EMA (PR #164): 32 -> 128 column buckets (256 regresses — most text/source lines fit in 128 columns, beyond which samples thin out). Wider column resolution separates e.g. mid-comment columns from code columns in source and deep-field columns in nci records. Gains led by dickens (89519->89500), nci (12997->12989), source (174888->174874). Local SCORE 568895 -> 568867 (-28). 9/9 round-trip lossless; only src/algorithm/ touched.

Validation

bash scripts/evaluate.sh passed locally; only src/algorithm/ changed.
Local SCORE: 568867 (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: 13516620133 (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).