docs: add full-repo repetition refactor audit proposal

[blackopsrepl]

Motivation

• Provide a methodical, workspace-wide audit to find repetition and high-impact refactor candidates across the Rust workspace.
• Focus on the largest crates (solverforge-solver, solverforge-scoring, solverforge-macros) where duplication most affects maintenance and LOC.
• Propose refactors while preserving performance constraints (zero-erasure, static dispatch) and keeping public APIs stable.

Description

• Add docs/refactor-audit-2026-04-20.md which documents the scan methodology, baseline LOC (57,387), six targeted refactor initiatives, and conservative LOC savings estimates.
• Enumerate concrete proposals for: arity consolidation in constraint streams, move-selector neighborhood unification, test harness extraction, manager/realtime scenario consolidation, scoring builder helper macros, and proc-macro codegen utilities, with an estimated aggregate savings of 4,370 LOC (~7.6%).
• Include a recommended execution order and guardrails (performance preservation, no runtime type-erasure, validate with make build/make test/benchmarks) to guide an incremental rollout.

Testing

• Ran formatting verification with cargo fmt --all -- --check, which completed successfully.
• No production code changes were introduced in this PR, so no crate build or unit-test runs were required as part of this change.

---

Codex Task

[blackopsrepl]

Closing in favor of the consolidated repetition-reduction plan in the repo-root PRD. This branch is based on v0.8.9 and is 19 commits behind current main, so it should be treated as input only rather than rebased forward.