feat(ruvector-core): Add OnnxEmbedding for real semantic embeddings

[ruvnet]

Summary

• Add native ONNX Runtime integration for production-ready semantic embeddings
• Implement OnnxEmbedding with auto-download from HuggingFace Hub
• Thread-safe inference via RwLock<Session>

Changes

File	Description
Cargo.toml	Add ort, tokenizers, hf-hub deps + onnx-embeddings feature
embeddings.rs	Implement OnnxEmbedding struct (~350 lines)
lib.rs	Export OnnxEmbedding, update deprecation warning
ADR-114	Update implementation status

Usage

use ruvector_core::embeddings::{EmbeddingProvider, OnnxEmbedding};

let provider = OnnxEmbedding::from_pretrained("sentence-transformers/all-MiniLM-L6-v2")?;
let embedding = provider.embed("hello world")?;
assert_eq!(embedding.len(), 384);

Test plan

• cargo check -p ruvector-core --features onnx-embeddings
• cargo test -p ruvector-core (123 tests pass)
• cargo clippy -p ruvector-core --features onnx-embeddings
• cargo publish --dry-run -p ruvector-core

Breaking Changes

None - This is an additive feature behind an optional feature flag.

Closes #264

🤖 Generated with claude-flow