[SPARK-40521][SQL] Return only exists partitions in PartitionsAlreadyExistException from Hive's create partition

[MaxGekk]

What changes were proposed in this pull request?

In the PR, I propose to modify Hive catalog implementation of create partition in the case when Hive reports that some of partitions exist already. In that case, Spark will check existence of all input partitions one-by-one and return only existing partitions in PartitionAlreadyExistsException.

Why are the changes needed?

1. To do not confuse Spark SQL users about existing partitions.
2. To be consistent with other catalogs V2 and V1 In-Memory.

Does this PR introduce any user-facing change?

Yes, it changes user-facing error message.

Before:

spark-sql> CREATE TABLE t (id bigint, data string) USING HIVE PARTITIONED BY (id);
spark-sql> ALTER TABLE t ADD PARTITION (id=2) LOCATION 'loc1';
spark-sql> ALTER TABLE t ADD PARTITION (id=1) LOCATION 'loc' PARTITION (id=2) LOCATION 'loc1';
The following partitions already exists in table 't' database 'default':
Map(id -> 1)
===
Map(id -> 2)

The error shows an existent and a non-existent partitions.

After:

...
spark-sql> ALTER TABLE t ADD PARTITION (id=1) LOCATION 'loc' PARTITION (id=2) LOCATION 'loc1';
The following partitions already exists in table 't' database 'default':
Map(id -> 2)

The error contains only existent partitions that cause issues.

How was this patch tested?

By running the modified test suites:

$ build/sbt -Phive-2.3 -Phive-thriftserver "test:testOnly *.AlterTableAddPartitionSuite"

[MaxGekk]

Here is another mistake. Must be id -> 2, I think. I will fix this separately as soon as I have this test in.

[MaxGekk]

Here is the fix: #38152

[MaxGekk]

@srielau @cloud-fan Could you review this PR, please.

[MaxGekk]

Merging to master. Thank you, @cloud-fan for review.