[P2][docs] Document performance envelope and bulk-load guidance

[cevheri]

Summary

The kernel copies the full sorted store per transaction (committed.slice() in transact). Bulk workloads are O(n²) in store size; entire DB lives in RAM. Seed/demo loops can look "hung" with no documented ceiling.

Audit ID: Section 2 Med — Wave 2 (documentation)

Fix (docs-first)

Add README section (or docs/PERFORMANCE.md if preferred):

• In-memory sorted array model → cost per commit scales with entry count.
• Guidance: wrap bulk loads in one transact(), not one per row.
• Honest ceilings: suitable for embedded/dev/test scale; not a server OLTP engine.
• Reference ARCHITECTURE.md for WAL replay cost on open.

Optional: mention #12 compaction for disk growth (not RAM).

Acceptance criteria

• Performance envelope documented where new users look (README or linked doc).
• Bulk-load anti-pattern called out explicitly.
• No changeset (docs only).

Agent constraints

• Do not promise future optimizations; document current behavior only.

Verification

Manual doc review.

[cevheri]

Implemented and shipped in PR #43 (merge commit 69f7161), released as v0.2.0. The PR description's comma-separated Closes list only auto-closed the first reference — closing manually. For #39/#28-style partials: the shipped scope is done; the real-browser verification half of #39 continues under #10, and the typed-error contract (#28) shipped as LibreDbError with stable codes across kernel, adapters, lenses, and CLI.