Parallel Tree Traversal and Distributed Lineage

Concepts: pyParallelDAG, pyKahns, pyWaveExecution

In a standard topological sort, you process nodes one at a time in order. But nodes at the same topological depth have no dependencies on each other — they can be processed in parallel. This is the key insight behind Spark's DAG scheduler, Airflow's task parallelism, and dbt's parallel model execution. The algorithm: compute all nodes with zero in-degree (no dependencies), execute them all in parallel, decrement the in-degree of their dependents, and repeat for the next wave. This is BFS-style topological traversal, but each wave runs in a parallel process pool. Level-Parallel Topological Execution Each 'wave' in this algorithm corresponds to one BFS level of the DAG. Nodes within a wave have no dependencies on each other, so they are safe to parallelize. Nodes in the next wave depend on n