feat(embed): embed CLI with download-only and JSON event output

[ohdearquant]

What

New embed CLI for the embeddings surface: --download-only, --json, --model, --text, --help.

Why

Lattice Studio's Embeddings screen needs a binary to fetch a model and produce embeddings. --download-only emits a @@lattice download-done event the app parses for progress. Genuinely new on main (the binary did not exist before).

Files

• crates/embed/src/bin/embed.rs (+201) — the CLI binary
• crates/embed/src/service/native.rs — adds NativeEmbeddingService::ensure_loaded (load-only path, no encode)
• crates/embed/Cargo.toml (+5: [[bin]] with required-features = ["native"])

Fix (codex round-1)

--download-only previously called service.embed(&["warmup"], model), running a full encode_batch after load. It now calls ensure_loaded(), which drives the same download + model-load path and returns without encoding — so the flag truly stops at load. Error handling and the JSON success event are unchanged; the embedding numeric path (SIMD / int8 / normalize / encode) is untouched, so the PR stays bench-neutral and the leaf-crate dependency direction (embed → inference) is intact.

Verification

cargo build --release -p lattice-embed --bin embed clean. cargo clippy -p lattice-embed --bin embed -- -D warnings clean. cargo test -p lattice-embed --lib green. Built in the integrated-tree gate (scripts/build-app-bins.sh). Codex round-2: APPROVE-WITH-SUGGESTIONS.

Bench

New binary; no existing library hot path modified. make bench-compare's comparator errored assembling the delta (known two-worktree fragility; base benches ran clean) — bench-neutral by construction, as the changed code is off every Criterion-measured path.

Series

Part of the PR #193 engine-slice (finest split). All engine code lands on main; the macOS app surfaces a subset (Models + Chat) for v0.0.1.

[github-actions]

E2E Parity Report

PASS: all 3 prompts match within first 3 tokens

Prompt	Agreement	First Diff	HF tok/s	Lattice tok/s	Verdict
The capital of France is	3/15	pos 3	0.4	2.6	PASS
In the year 2024, artificial intelligence	10/15	pos 9	0.4	2.2	PASS
`def fibonacci(n):

if n <= 1:
    return n
return` | 15/15 | none | 0.4 | 1.9 | PASS |

The capital of France is

• HF: Paris.
  The capital of France is Paris.
  The capital of France
• Lattice: Paris.
  A: Yes, the capital of France is Paris.

In the year 2024, artificial intelligence

• HF: (AI) has become a significant part of the global economy. It is
• Lattice: (AI) has become a significant part of our daily lives. From personal

def fibonacci(n): if n <= 1: return n return

• HF: fibonacci(n-1) + fibonacci(n-2)

print(fib

• Lattice: fibonacci(n-1) + fibonacci(n-2)

print(fib