docs: Rust core port plan — decompose #6 into actionable sub-issues

[LeslieOA]

Summary

Adds docs/rust-core-port-plan.md (~250 lines). Converts #6 from a single XL block into a structured plan with six open architectural questions for you to answer, and a proposed decomposition into 14 sub-issues that get filed once those questions are answered.

This PR is not a commitment to start coding. It's a conversation starter that makes #6 reviewable, decision-able, and ready to chunk into parallel work.

Why this is the right next step

Everything we shipped in #2 (port) / #3 (renderer port) / #4 (history refactor) / #5 (Partial → variants) was Rust-port prep. The API is now FFI-friendly. But #6 itself has been "XL, undefined" — anyone picking it up has to make decisions about UniFFI vs JSI, rollout shape, build integration, and testing strategy before writing any Rust. This doc surfaces those decisions explicitly and proposes answers.

The six open questions

#	Question	Recommendation
1	Rollout: big-bang vs parallel-with-flag vs per-function migration?	Parallel with flag, 6-month overlap
2	Native FFI tooling: UniFFI vs JSI direct vs Turbo Modules?	UniFFI + JSI direct as escape valve
3	Web WASM loading: separate file vs inline base64?	Separate file (matches Skia pattern)
4	Scope creep — port markdown rendering too?	No (out of scope; would need comrak / pulldown-cmark or port unified/remark)
5	Timing — start soon or sit while we prove out more renderer / harness work?	Your call
6	Rust expertise — internal or external?	Your call

Proposed sub-issue decomposition

14 sub-issues, 2 L / 5 M / 7 S, rough dependency order. Most can run in parallel. Realistic delivery: 4-8 weeks elapsed with focused part-time work.

#	Title	Size
1	chore(rust): set up Cargo workspace + jsoncanvas-core crate skeleton	S
2	feat(rust): port types.ts to Rust structs/enums with serde	S
3	feat(rust): port parseCanvas / serializeCanvas	M
4	feat(rust): port spatial-index	M
5	feat(rust): port operations + canvas-state + history	L
6	chore(rust): UniFFI interface + bindgen for iOS	M
7	chore(rust): UniFFI bindgen for Android + macOS	S
8	chore(rust): wasm-bindgen integration for web	M
9	feat(rust): TS facade layer	S
10	chore(rust): Expo config plugin for native builds	M
11	test(rust): cross-impl conformance test runner	S
12	test(rust): property-based round-trip tests	S
13	feat(rust): opt-in flag for Rust impl (default off)	S
14	docs: migration guide for consumers	S

What's deliberately NOT in this plan

Surfaced in the doc itself (§10) but worth flagging: a bytecode cache for parsed canvases, streaming parse, persistent / mmap'd storage, multi-canvas state. All real ideas; none belong in #6. Each could be its own follow-up after Rust ships.

How to use this PR

1. Read the doc (it's 250 lines, mostly structured tables and lists)
2. Answer the six open questions (in PR review, on feat(core): port pure-logic core to Rust (JSI native + wasm-bindgen web) #6, or here)
3. I file the 14 sub-issues, update feat(core): port pure-logic core to Rust (JSI native + wasm-bindgen web) #6's body to point at them as an umbrella
4. Sub-issues get prioritised and picked up independently

Verification

• npm test: 38/38 (develop hasn't merged the conformance tests from test: conformance tests for JSON Canvas spec + Canvas Candy baseline #29 yet)
• npm run typecheck: clean
• npm run lint: exit 0
• No code changes; pure docs

Refs: #6