v0.0.7

Added

• Local embeddings pipeline with fastembed v5 and AllMiniLML6V2 model
• Embedder struct with automatic model download and caching to ~/.idep/models/
• Embedder::embed_batch() method with 384-dimension assertion for AllMiniLML6V2
• EmbedPipeline for batch processing with configurable batch size (default 32)
• EmbeddedChunk struct combining CodeChunk with Vec embedding
• Progress callbacks for large batch embedding operations via run_with_progress()
• Comprehensive test suite:
  • Unit test: embed batch of 10 strings, verify 384-dimension shape
  • Benchmark: time to embed 100 chunks of ~200 tokens each
  • Network independence test: verifies no outbound network calls after initial download
  • Pipeline tests: verifies one embedding per chunk, custom batch sizes, progress callbacks
• Model cache exclusion from git via .gitignore (.fastembed_cache/, *.onnx, *.lock)
• Internal workspace dependencies for cross-crate integration

Changed

• Added PartialEq to ChunkKind for testing pipeline functionality
• Workspace version bumped to 0.0.7
• All 62 tests passing across all crates