ragleap-vectorstores v0.3.0 release notes

Added

RedisBackend - third vector backend, via RediSearch/Redis Stack's HNSW vector index. Available via the redis optional extra (pip install ragleap-vectorstores[redis]).
Requires a real running server (Redis Stack, or plain Redis + the RediSearch module) - unlike the embedded Chroma/LanceDB backends. init_schema() checks for the module and raises a clear error if it's missing.
Honest limitation: metadata filtering only supports document_id (RediSearch requires predeclared schema fields, unlike Chroma's flexible where=).
supports_sparse() reports False - RediSearch does support real full-text search, but that surface wasn't implemented/tested in this release.

TAG field values (e.g. UUID document_ids) are now properly escaped in RediSearch queries - an unescaped hyphenated ID silently returned zero matches.

15 new tests, 40/40 total passing. See CHANGELOG for full details.