v0.1.0

feat: add chunks-to-pgvector skill and embed_and_store_chunks MCP tool

New capability: export PDF chunk embeddings to PostgreSQL with pgvector.

- src/malimgraph/core/embedder.py — multi-provider embedding generation
  (OpenAI text-embedding-3-*, Voyage voyage-3-*, local sentence-transformers)
- src/malimgraph/core/vector_client.py — PgVectorClient with HNSW index,
  upsert, cosine similarity search, stats, list, delete
- server.py — new MCP tool: embed_and_store_chunks
- cli.py — new CLI group: malimgraph vector load/search/stats/list/delete
- skills/chunks-to-pgvector/ — full Claude skill with SKILL.md + 3 scripts:
  embed_chunks.py, search_vectors.py, manage_vectors.py (all with inline fallbacks)
- pyproject.toml — new optional deps: pgvector, openai, voyage, local
- Makefile + publish.yml — include chunks-to-pgvector.skill in release assets
- tests/test_vector_client.py — unit tests for embedder + client

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>