deploy: implement Core ML export

[AbdelStark]

Context

Primary deployment target for Apple Silicon.

Scope

• Implement geno_lewm/deploy/coreml.py::export.
• Use coremltools with the Neural Engine target where available.

Out of Scope

• iOS-specific bundling.

Design Reference

• RFC: rfcs/0010-on-device-personal-genome-deployment.md §3.2

Acceptance Criteria

• M3 Max single-variant warm latency under target (RFC-0007 §3.3)
• Numerical drift within quantization budget

Parent tracking issue: #14

[AbdelStark]

Blocked (session-scoped): requires runtime toolchains not present in this environment.

This issue depends on ONNX / Core ML / GGUF / quantization tooling whose installation, version pinning, and host-OS support is non-trivial:

• ONNX Runtime + onnx-converter for the export path.
• Core ML Tools (macOS-only conversion path).
• llama.cpp's convert.py family for GGUF.
• bitsandbytes / optimum-quanto for int8/int4 calibration.

Beyond installation, the issue requires a trained predictor (predictor.safetensors) to convert — which itself depends on #41 (predictor) and #44 (trainer) landing.

Foundations now in place that the export pipeline will plug into:

• geno_lewm.errors.ExportFormatError / QuantizationError / BackendUnsupportedError for typed failures.
• geno_lewm.provenance.{sha256_file,Manifest} for the post-export model_id derivation that the runtime loader (#66) will compare against.

To unblock: install the target toolchain(s), wait for #41 / #44 to ship a checkpoint, and pick this up.

[AbdelStark]

Blocked in the current repo state.

Core ML export depends on the same missing checkpoint contract as #67 plus Core ML-specific tooling and hardware validation:

• RFC-0010 §3.2 defines export as converting a trained predictor + action_encoder checkpoint into deploy artifacts. The repository still has no implemented checkpoint producer/loader for predictor.safetensors / action_encoder.safetensors.
• training: complete clean-machine Carbon-backed trainer run #44 (training: implement trainer scaffold) remains open and labeled status:blocked, so there is no trainer-owned trained checkpoint to convert.
• geno_lewm/cli/export.py is still a Phase 1 stub pending deploy: implement geno-lewm-export CLI #71 on top of the per-target export modules.
• Current toolchain probe: platform: Darwin/arm64, but torch: False, coremltools: False, safetensors: False in the dev environment.
• Acceptance requires M3 Max single-variant warm latency and quantization-drift validation; that cannot be satisfied without the Core ML toolchain, target hardware benchmark setup, and a real checkpoint.

Unblock once the checkpoint schema/producer lands, Core ML tooling is pinned in a supported environment, and the M3 Max benchmark fixture is available. Then implement geno_lewm/deploy/coreml.py::export against the real artifacts and validate latency/drift.