pub fn encode_arc_records_partitioned<T: Serialize + Sync + Send>(
schema: &Schema,
records: &[Arc<T>],
) -> Result<Vec<Vec<u8>>, KineticaError>Encodes Arc-wrapped records, partitioning the batch over threads.
This strategy partitions the batch into chunks (one per thread) and processes each chunk sequentially within the thread. This reduces:
- Lock contention (each thread has its own chunk)
- SerializerConfig creation overhead (reused within chunks)
- Memory allocation (pre-allocated buffers per chunk)