[SPARK-4648][SQL] Support COALESCE function in Spark SQL and Hive QL

[ravipesala]

Support Coalesce function in Spark SQL.
Support type widening in Coalesce function.
And replace Coalesce UDF in Spark Hive with local Coalesce function since it is memory efficient and faster.

[SparkQA]

Test build #23938 has started for PR 3510 at commit bbdeebe.

• This patch merges cleanly.

[SparkQA]

Test build #23938 has finished for PR 3510 at commit bbdeebe.

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

[AmplabJenkins]

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/23938/
Test FAILed.

[SparkQA]

Test build #23942 has started for PR 3510 at commit 9fa1f9e.

• This patch merges cleanly.

[AmplabJenkins]

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/23940/
Test FAILed.

[ravipesala]

retest please

[SparkQA]

Test build #23942 has finished for PR 3510 at commit 9fa1f9e.

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

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/23942/
Test PASSed.

[SparkQA]

Test build #23949 has started for PR 3510 at commit e377a0c.

• This patch merges cleanly.

[SparkQA]

Test build #23949 has finished for PR 3510 at commit e377a0c.

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

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/23949/
Test PASSed.

[chenghao-intel]

Nit: Move that into the pattern matching guard?

[marmbrus]

Though then you would have to do the map/distinct twice I believe.

[chenghao-intel]

That's a good catch, but the children data type converting criteria for Coalesce need to be discussed more deeply. Generally, I think we'd better keep it the same as Hive does. And it will be nicer if a Hive query suite provided.

[marmbrus]

indentation is off here.

[marmbrus]

Thanks for working on this! In order to keep the PR queue small, I propose we close this issue for now and reopen it once we figure out the correct coercion semantics and add more tests.