get
Most-recently-processed chunk for the source. A bare @NodeFragment has no generated query DSL, so the "latest by processedAt" pick is a small parameterized Cypher returning the id; the node itself is then loaded via the high-level API (correct temporal deserialization).