pgvector 0.6.0: 30x faster with parallel index builds

The latest release of pgvector (0.6.0) introduces significant improvements in HNSW index building speed. Parallel builds for HNSW indexes have made the process up to 30 times faster, particularly for unlogged tables. This enhancement is expected to improve search accuracy and performance while tuning HNSW build parameters. The benchmark results show that the new version can build an index for a million vectors in just 9.5 minutes, compared to over an hour with the previous version. Additionally, using unlogged tables can further reduce building time by up to 31 times.