qdrant_vectors.py demonstrates batch CRUD, named dense-vector selection,
native portable filtering, and score thresholds using deterministic vectors.
It creates and deletes its own unique collection.
Run a disposable Qdrant 1.16.2+ server, set QDRANT_URL, and follow the
package README. No embedding
service is needed. Optional QDRANT_API_KEY is supported by the sample and
the connector's QdrantSettings, which also supports explicitly selected .env files.
Portable filters deliberately require a server, not :memory: local mode.