How does the

SnowflakePandasTypeHandler

works if we add it to an Op with that doesn’t have partitions? Does the table get truncated on each run? (Is there a way to make it append only instead?)