[SPARK-32704][SQL] Logging plan changes for execution
### What changes were proposed in this pull request?
Since we only log plan changes for analyzer/optimizer now, this PR intends to add code to log plan changes in the preparation phase in `QueryExecution` for execution.
```
scala> spark.sql("SET spark.sql.optimizer.planChangeLog.level=WARN")
scala> spark.range(10).groupBy("id").count().queryExecution.executedPlan
...
20/08/26 09:32:36 WARN PlanChangeLogger:
=== Applying Rule org.apache.spark.sql.execution.CollapseCodegenStages ===
!HashAggregate(keys=[id#19L], functions=[count(1)], output=[id#19L, count#23L]) *(1) HashAggregate(keys=[id#19L], functions=[count(1)], output=[id#19L, count#23L])
!+- HashAggregate(keys=[id#19L], functions=[partial_count(1)], output=[id#19L, count#27L]) +- *(1) HashAggregate(keys=[id#19L], functions=[partial_count(1)], output=[id#19L, count#27L])
! +- Range (0, 10, step=1, splits=4) +- *(1) Range (0, 10, step=1, splits=4)
20/08/26 09:32:36 WARN PlanChangeLogger:
=== Result of Batch Preparations ===
!HashAggregate(keys=[id#19L], functions=[count(1)], output=[id#19L, count#23L]) *(1) HashAggregate(keys=[id#19L], functions=[count(1)], output=[id#19L, count#23L])
!+- HashAggregate(keys=[id#19L], functions=[partial_count(1)], output=[id#19L, count#27L]) +- *(1) HashAggregate(keys=[id#19L], functions=[partial_count(1)], output=[id#19L, count#27L])
! +- Range (0, 10, step=1, splits=4) +- *(1) Range (0, 10, step=1, splits=4)
```
### Why are the changes needed?
Easy debugging for executed plans
### Does this PR introduce _any_ user-facing change?
No.
### How was this patch tested?
Added unit tests.
Closes #29544 from maropu/PlanLoggingInPreparations.
Authored-by: Takeshi Yamamuro <yamamuro@apache.org>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
0cb91b8c18