[M4n5ter]

Summary

Synchronize the complete vendored NSH tree with longbridge/nsh main at a5036af48ddc193b5023ece9b9196c76b8f8d3f5, consolidating previous selective imports into one recorded upstream baseline. Most of the 912-file diff comes from upstream source, components, tests and documentation; Maka workbench consumer changes are confined to two files.

Migrate consumers to the current APIs (ItemAlign, platform.desktop, and listener stop functions), while retaining TUI, input/IME ownership, dynamic module loading, and list publication/retirement behavior. Adopt the shared image loader with isolated inline-byte cache keys and stable asset-reader identities across reloads. Refresh provenance from original upstream bytes and document future synchronization in third-party/nsh/SOURCE.md.

Keep image reads queued without adopting upstream's 512-entry rejection, and preserve unrestricted local-file reads. HTTP downloads retain their existing 64 MiB boundary; the new image-clipboard API retains its 48 MiB boundary for the finite JS heap. Replace the arbitrary mount-operation threshold with a 5,000/50,000-item scaling comparison, and reuse upstream's pointer/keyboard behavior coverage.

Verification

• NSH core, base, engine, JS, headless, TUI, bundler and development tooling: 3,511 passed, 0 failed, with 41 existing ignored tests/examples.
• GPUI owning workspace: 24 image tests passed. Portable Web text tests: 31 passed; Web runtime JavaScript tests: 27 passed.
• Fresh NSH/Host declarations pass the real workbench TypeScript check; SDK/product type checks and CLI/desktop compilation checks pass.
• Strict Clippy passes for the affected NSH crates and Maka workbench/CLI/desktop targets. Rust formatting, workbench Biome checks and warning-free core rustdoc pass.
• Source headers, dependency inventory, regenerated notices and their regression test pass. Product packaging tests pass.

Local validation covered Linux compilation and headless/unit behavior. Native window interaction, browser rendering and other operating-system runtimes were not exercised locally.

AI use

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope: Codex assisted with upstream integration, API migrations, regression coverage, provenance and documentation. The commit carries Generated-by: Codex.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No