feat(search): support distributed batch vector queries

[ddupg]

Summary

• support fixed-size vector query batches with per-query global top-k across indexed and flat-fallback fragments
• preserve Lance projection and distance semantics across partial index coverage, including index metric inheritance, bit-level Hamming distance, distance ranges, and requested row IDs
• document the batch result contract and cover IVF_FLAT, IVF_SQ, IVF_PQ, partial-index, flat-fallback, and multivector boundaries

Testing

• pytest tests/test_distributed_search.py (37 passed)
• pytest tests/test_distributed_indexing.py::test_build_distributed_vector_index (3 passed)
• ruff format --check lance_ray/search.py tests/test_distributed_search.py tests/test_distributed_indexing.py
• ruff check lance_ray/search.py tests/test_distributed_search.py tests/test_distributed_indexing.py
• git diff --check