docs(adr-001): README complexity-class surface reflects shipped phase-2

[ruvnet]

Summary

Closes the README half of the ADR-001 SOTA polish. The "Complexity as a First-Class API Surface" section was stale:

• `find_anomalous_rows` marked "Linear today, → SubLinear phase 2" — but the phase-2 path (`contrastive_solve_on_change_sublinear`) shipped in PR feat(adr-001 #6): phase-2B — single-entry Neumann + SubLinear orchestrator #27.
• The new SubLinear primitives — `closure_indices`, `solve_single_entry_neumann`, `solve_on_change_sublinear`, `contrastive_solve_on_change_sublinear` — weren't listed at all.
• No empirical receipt was given for the SubLinear claim.

Changes

• Updated table to surface every shipped SubLinear primitive with the right class label.
• Added bench-receipt numbers from PR bench(adr-001 #2/#6): empirical SubLinear delta-solve comparison #31's `delta_solve` group (cold_full / warm_full / sparse_closure across n = 64, 256, 1024).
• Clarified MCP enforcement coverage (`solve`, `estimateEntry`, `solveTrueSublinear`).
• Listed every phase-2 method name now queryable via `estimateComplexityClass`.

Status of ADR-001 SOTA gaps

• README polish (this PR)
• BENCHMARK.md update (deferred)
• Pi hwmon J/solve backend (deferred — needs hardware runner)

Test plan

• Markdown renders correctly in GitHub preview
• CI (no Rust/TS code touched; only docs)

🤖 Generated with claude-flow