[SPARK-15970][SQL]: avoid warning message for hive metastore in In-memory ca…

[xwu0226]

What changes were proposed in this pull request?

Issue:

Switch the catalog implementation mode to in-memory mode by adding spark.sql.catalogImplementation in-memory in spark-defaults.conf.
Then, start spark-shell, which will be run using in-memory catalog.
Issue DDL command like CREATE TABLE T1 (c1 INT, c2 INT) USING PARQUET PARTITIONED BY (c2). The table metadata will be created in in-memory catalog. However, a WARNING message dumped into the console says that Persisting partitioned data source relation default.t1 into Hive metastore in Spark SQL specific format, which is NOT compatible with Hive. This kind of warning message is not applicable to In-Memory catalog mode.

This PR is to skip checking whether the table can be created as hive compatible table or not when in-memory catalog mode is in use.

How was this patch tested?

A unit test case is added for in-memory catalog mode. And regtest buckets are run locally.

[SparkQA]

Test build #60588 has finished for PR 13687 at commit f6b1717.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[SparkQA]

Test build #62726 has finished for PR 13687 at commit f6b1717.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[SparkQA]

Test build #69748 has finished for PR 13687 at commit f6b1717.

• This patch passes all tests.
• This patch does not merge cleanly.
• This patch adds no public classes.

[gatorsmile]

@xwu0226 We did a major refactoring in the codes. Do we still face the same issue? If not, maybe we can close it?

[xwu0226]

Let me rebase and code and try again . Will close it if we don't have any issue. Thanks!

[xwu0226]

Closed since the hive refactoring code has resolve this issue by the way.