papers: design sketch outline for arxiv position paper (#67)

[proofmancer]

First scaffolding for issue #67 (the design-sketch / position paper). Lands a `papers/` directory with its own lifecycle (outline -> draft -> review -> published -> superseded) and a 12-section outline for `design-sketch.md` that fixes the paper's structure before any prose work begins.

What landed

`papers/` directory + lifecycle

• `papers/README.md` documents the audience (researchers, engineers at competing chains, funders), the distinction from `spec/rfcs/` (proposals) and `spec/protocols/` (stable specs), and the lifecycle.
• `papers/design-sketch.md` with `status: outline` frontmatter.

Paper outline: 12 sections + appendices

1. Introduction
2. Background: what's broken about today's chain languages
3. The chain manifest as the central abstraction
4. Multi-VM execution
5. Effect-typed contracts
6. Multi-dimensional gas
7. Sampling-first data availability
8. What we deliberately don't optimize
9. Implementation status
10. Roadmap
11. Related work
12. Conclusion

Plus appendices: grammar summary, selected example modules, RFC index.

Each section has:

• 2-5 bullets sketching what the final prose will argue
• Citations the section will assemble
• Open questions for the drafting phase

Drafting plan

• Lock outline (this PR) — done
• Draft section by section over 3-4 weeks, 1 PR per section
• Internal review pass
• External review pass (target: Sui Labs / Move project, Substrate background, academic)
• Final polish + publish to arxiv + project website

Target: 10-15 pages typeset, draft in ~3 months from outline.

Why an outline first

Three reasons:

1. Structure matters more than prose. A paper with a confused structure stays confused; one with a clear structure can have its prose iterated. Locking the section list first means subsequent PRs are focused.
2. Forces scope discipline. Without the outline, drafting would creep into "well, also let me argue X." With it, additions are explicit deviations that can be debated.
3. Enables parallel drafting. Multiple contributors (when we have them) can draft different sections without stepping on each other.

What this PR does NOT do

• Not a draft. Every section is bullet points + drafting notes. No prose to ship.
• Not a commitment to publish. Publishing is downstream; if the outline reveals the paper doesn't make sense, we'll fold it back into an RFC instead.
• Not the empirical paper. This is the position paper / design sketch. The empirical paper (real chain throughput vs Substrate / Cosmos / Move) lands ~12-18 months from now, after the chain layer ships.

What unblocks next

• Drafting can start section by section (suggested order: §3 chain manifest -> §4 multi-VM -> §6 multi-dim gas -> §7 sampling-first DA -> §5 effects -> §2 background -> §9 status -> §1 intro -> §12 conclusion -> §10 roadmap -> §11 related work -> §8 non-goals). Section order in the paper is fixed; drafting order is whichever has the clearest argument first.
• Anyone reading the repo can now see how Cleave's design fits together at a higher level than the RFC structure.
• The outline doubles as a checklist for what RFCs and code links need to be ready by paper-draft time.