[SPARK-9050] [SQL] Remove unused newOrdering argument from Exchange (cleanup after SPARK-8317)

[JoshRosen]

SPARK-8317 changed the SQL Exchange operator so that it no longer pushed sorting into Spark's shuffle layer, a change which allowed more efficient SQL-specific sorters to be used.

This patch performs some leftover cleanup based on those changes:

• Exchange's constructor should no longer accept a newOrdering since it's no longer used and no longer works as expected.
• addOperatorsIfNecessary looked at shuffle input's output ordering to decide whether to sort, but this is the wrong node to be examining: it needs to look at whether the post-shuffle node has the right ordering, since shuffling will not preserve row orderings. Thanks to @davies for spotting this.

[JoshRosen]

/cc @yhuai and @davies for review; this should be a pretty straightforward cleanup patch.

[yhuai]

LGTM

[davies]

LGTM

[SparkQA]

Test build #37283 has finished for PR 7407 at commit e70be50.

• This patch fails Spark unit tests.
• This patch merges cleanly.
• This patch adds no public classes.

[SparkQA]

Test build #37281 has finished for PR 7407 at commit e866494.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • case class Exchange(newPartitioning: Partitioning, child: SparkPlan) extends UnaryNode

[JoshRosen]

Alright, I'm going to merge this into master. Thanks for looking this over!