docs: explain why ilo has no borrow types

[danieljohnmorris]

Summary

Brief mem/4: add a defensive doc section to zero-gap-specs/lessons-from-zero.md explaining why ilo deliberately has no user-visible borrow type annotations (ref<T>, mutref<T>, owned<T>). Stops future contributors from re-deriving the rationale every time the question comes up, and gives potential users a clear signal that this is a deliberate design choice with tradeoffs we can articulate, not an oversight.

Note: zero-gap-specs/ was previously kept locally outside the repo. This PR commits the lessons-from-zero doc into the ilo repo so the planning rationale is version-controlled alongside the code it justifies. The expanded section 5a is the substantive change; the rest of the file is the existing planning doc copied as-is so future edits flow through normal PR review.

What's in the diff

• Expanded ### a) Borrow checker / ownership types in section 5 from a one-line dismissal to a full rationale citing Principles 1 (Token-Conservative) and 4 (Language-Agnostic) by name, plus the "one layer of the stack" framing from section 4h and the manifesto.
• Added section 10e (escape analysis and linearity inference for RC elision) so the forward-link from section 5a has something to point at. Renumbered the old 10e (multi-agent orchestration) to 10f.
• Honest about the tradeoff: RC has runtime overhead vs static ownership, and that overhead is the cost agents pay in exchange for source-token simplicity. Generous to Rust and Zero - borrow types are correct for their audiences, wrong answer for ilo's.

Test plan

Docs-only change. No code paths touched.

• New section references Principles 1 and 4 by name
• Forward-link to section 10 (escape analysis) resolves to a real entry
• Tone is generous to Zero and Rust, framed as "different audience, different choice"
• Under 250 words for the new prose block

Follow-ups

• Public-facing memory page (separate brief, mem/5).

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ All tests successful. No failed tests found.

📢 Thoughts on this report? Let us know!