Migrate from AnnotateWithSequenceNumber to AttachDistributedSequence

[`AttachDistributedSequence`](https://github.com/apache/spark/blob/master/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/plans/logical/pythonLogicalOperators.scala#L367) seems to do more/less what `AnnotateWithSeqenceNumber` does, and can probably do it FAR FAR faste.  