[SPARK-16646][SQL] LEAST and GREATEST doesn't accept numeric arguments with different data types

[HyukjinKwon]

What changes were proposed in this pull request?

This PR makes LEAST and GREATEST accept other numeric types (Decimal). This worked fine in HiveContext in 1.6 because , for example, 1.5 was DoubleType in there.

So, this finds a tightest common type from both IntegerType and DoubleType as DoubleType, then it works okay.

But currently it seems 1.5 is being treated as DecimalType for now. So, It casts 1.5 as decimal(2, 1). So, it fails to find a tightest common type from both IntegerType and DecimalType(2, 1).

This PR introduces new function findTightestCommonTypeToDecimal dealing with decimals interacting with each other or with primitive types. This logic is borrowed from JSON schema inference.

How was this patch tested?

Unit tests in TypeCoercionSuite and DataFrameFunctionsSuite.

[HyukjinKwon]

@liancheng and @davies Will this change be appropriate? Could you please take a look?

[SparkQA]

Test build #62652 has finished for PR 14294 at commit 9730afb.

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

[SparkQA]

Test build #62653 has finished for PR 14294 at commit abbcb4e.

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

[HyukjinKwon]

retest this please

[SparkQA]

Test build #62657 has finished for PR 14294 at commit abbcb4e.

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

[HyukjinKwon]

Closing this. Please refer https://issues.apache.org/jira/browse/SPARK-16646